Quality and integrity in the translation of official documents

ABSTRACT

The translation of official personal documents is widely needed and practised. As people migrate physically or virtually for employment and other purposes, millions of personal documents are translated and submitted to public administrations and private companies. This is also one of the areas of translation that is most regulated around the world, with translators often required to sit a test, demonstrate relevant qualifications, and/or be sworn in by the relevant ministry or administration. However, the confidential nature of the source documents often implies that the quality of the corresponding translations can only be assessed by the clients themselves or the institutional staff who process the applications and cases. In this paper, we assess the quality and integrity features of a corpus of translated official personal documents collected in Australia. While the analysis shows a high standard overall, it also reveals areas for improvement.

KEYWORDS:

• Translation
• official documents
• assessment
• quality
• translation integrity

Introduction

As Taibi and Ozolins (2016) have noted, the translation of official personal documents is generally treated as being part of the expanding field of community translation. However, when we consider that official document translation plays a pivotal role wherever we find the movement of people (and corresponding need to process identity, qualifications and other documents), its salience as a specialisation is undeniable – and moreover increasingly acknowledged, as witnessed by EU efforts to introduce common standards for the practice (European Union, 2016).

The usual paradigm in terms of ‘community translation’ is the material published by governmental or NGO bodies for the information needs of minority groups, and the corresponding issue of conveying sometimes unfamiliar processes and information to the target readership through techniques such as explication, explanation and appropriate communicative adjustment; this is certainly where theory and practice have applied their main focus.

Contrasted against community translation, official document translation shows some clear and significant differences, not least being the international rather than local element at its core. It deals with individuals from a potential multitude of countries, and their data need to be put into a comprehensible form to enable the host society’s institutions to make judgments about the identity, status or qualifications of (usually) new arrivals – for whom establishing one’s bona fides is clearly an important step in the settlement process, and can be key to accessing high-stakes opportunities. Translations of official documents are therefore produced under tight and exacting constraints to ensure the correct rendering of identity, status and qualifications documents that have been issued, obtained and recorded elsewhere; in turn, the institutions that receive and rely on them play an essential gatekeeper role.

When it comes to what constitutes quality in official document translations, little of the literature on translation quality has paid attention to the genre. Official translations are normally commissioned by or on behalf of private clients and processed internally by administrative or institutional bodies; as such they stand somewhat apart, and many would not appear to undergo the same quality assurance mechanisms that are now commonly implemented by individual translators or translation companies engaged in other areas. In addition, unlike various other types of translation (e.g., literary, scientific, audiovisual, websites, etc.), translations of official personal documents are subject to strict confidentiality, and therefore remain beyond the reach of public or expert scrutiny. These factors make it very difficult under normal circumstances to assess official translation quality and identify any issues.

With that in view, our purpose in this article is precisely to run a quality assessment analysis on a corpus of official document translations in order to identify strengths and weaknesses in the practices of Australian translators working in this area of translation. This article arises from a broader study of official document translations involving a survey of translators, language service providers (LSPs) and end users reported elsewhere (Taibi & Ozolins, forthcoming a, forthcoming b) while this article deals specifically with quality assessment. The analysis is intended to suggest improvements that can be made by translators or translation companies to ensure clients and institutions receive quality translations of these often ‘make-or-break’ documents. We also hope to contribute to refining theoretical approaches to quality in this specific area of translation practice.

Translation quality

To frame our discussion, we use the pathbreaking work of Drugan (2013, 2018), which outlined two historic approaches to translation quality. The first, often dubbed ‘academic’, asks what features of a completed translation can be assessed as providing quality, and by what criteria – linguistic, contextual, purpose-related – can such quality be measured. Broadly grouped under the rubric of Translation Assessment [TA], authors such as House (1977, 1997, 2014) and (Nord, 1997/2018) figure prominently.

By contrast, Drugan points out that in the professional field of translation, quality is viewed in a quite different and – for both translators and language service providers (LSPs) – much more problematic way: that is, what measures do they put in place to assure the quality of their product to their clients before clients entrust them with their translations? This perspective, encapsulated under the rubric of Translation Quality Assurance (TQA or often simply QA), has both marketing and reputational implications for translators. And rather than look at translation post facto, it goes to the heart of the trust between client and translator (or translation company) in negotiating translation assignments:

Translation quality matters in the industry, and for different reasons than in translation studies. Providers have to measure, compare and guarantee quality throughout the translation process. Before winning contracts, they must convince clients they can deliver translations more reliably or efficiently than rivals. (Drugan, 2018, p. 1)

This is not to say the traditional or academic work on translation quality is to be ignored. Significantly, even in the academic literature on translation quality, important theoretical understandings of how official translations need to be viewed can be obtained, despite the heavy focus of translation theorists on literary translation.

We lead our present survey of the literature with the insightful work of Nord (1997/2018), whose view of documentary versus instrumental translation we consider the closest to a theoretical understanding of how quality can be conceptualised for official documents.

In Nord’s theoretical view, documentary translation

aims at producing in the target language a kind of document of (certain aspects of) a communicative interaction in which a source-culture sender communicates with a source-culture audience via the source text under source-culture conditions. (1997/2018, p. 46)

More specifically, Nord argues that a documentary translation is

a text whose main function is metatextual […] The target text, in this case, is a text about a text, or about one or more particular aspects of a text. (1997/2018, p. 46)

As opposed to this, Nord (1997/2018, p. 47) categorises instrumental translations as ‘producing in the target language an instrument for a new communicative interaction between the source-culture sender and a target-culture audience, using (certain aspects of) the source text as a model’. In particular, an instrumental translation does not necessarily announce itself as a translation but can play the function of a natural text in the target language. A motor car manual may well be a translation, but the fact of its being a translation would not necessarily be obvious or even of interest to a target-language reader. Similarly, novels can be read without consciousness of their being a translation.

Similar distinctions have been made by other authors, including House (1977, 2014) who identified overt and covert translation types – an overt translation clearly identifies itself as a translation (a point to which we will return); covert translations try to represent themselves as a normal text in the target language, which may not be even recognised as a translation by the reader. Thus, in literature or films, the Aladdin of medieval Arabia is transformed to become a boy with many contemporary western features.

Now, Nord was principally interested in styles of literary translation, and certainly did not develop her theory of documentary translation specifically to cover official documents. However, these clearly do fit into her definition of documentary translation, as among the examples she gives are ‘certificates’. Nord’s concern for functionalism and skopos also fits in well in identifying the purpose of official document translation, and she gives analogies from her literary examples – amongst which she cites the close textual reading and continual annotation that marks the work of those trying to decipher and translate original classical religious texts, where the awareness of source text features and mirroring source text vocabulary and even syntax is a continual process of uncovering meaning and making constrained choices in translation.

On a different level, Toury’s (2012) well-known distinction between adequacy and acceptability and Venuti’s (2018) distinction between foreignisation and domestication as basic approaches in translation can also give something of the flavour of official documents. While domesticating approaches prioritise target culture norms (as in the Aladdin example above), a foreignising approach brings in elements which are perhaps unfamiliar to the target-language audience and alerts the reader to the need to understand them in terms of their native cultural milieu; in the case of official documents, this becomes a need for a relentless awareness of how both the texts and their contexts reflect (and are embedded in) a different and particular institutional and societal framework.

For the functionalist approaches of Nord and House then, measuring the quality of translations relates primarily to how well they fulfil their diverse functions, whatever they may be. These approaches have consolidated around the theory of skopos, or purpose-specific translation, with a particular emphasis on register and target audience to fulfil various intended functions.

Here House departs from her other functionalist colleagues, arguing that the focus on skopos – understood as purpose – gives too vague an understanding of possible quality elements. In her look at the history of quality assessment and her own theoretical summary (House, 2014), she argues that a concern for equivalence must underlie all translation assessments, but that this is not a simple literal equivalence. House argues that an approach based on the Hallidayan notion of register can build a profile of a source text which can be differentiated and specified by reference to the established categories of field (the subject matter), mode (written/spoken/ interactive/static) and tenor (relationship between author and receiver). To make the specificity more accurate, the notion of genre is also introduced, with a critical requirement that the genre be reproduced in the translation. By constructing such a source text profile and using it to determine which elements show equivalence, we have effectively performed an analysis of quality.

House introduces one other theoretical concept which also brings her work in line with that of Nord: House differentiates translations as to whether they are overt (i.e., announcing themselves as translation) or covert (where reader comprehension does not necessarily depend on knowing it is a translation). Relating this to official document translation, House and Nord both bring us to an appreciation of the linguistic features of official translations and their specific register and genre.

Official translations are clearly overt, documentary translations, reflecting a specific institutional setting (field) in usually highly constrained written form (mode) and addressed to any reader needing to deal with the holder of the document (tenor) – we will discuss this further below. The translation then reproduces these features with the difference that, qua translation, it addresses readers in a different institutional setting (the host society). How then can this best be characterised? Despite House’s scepticism of the theoretical power of skopos approaches, we find in this writing a key insight, that the translation is very much, as Reiss (2014) argues, an offer of information to the reader in a new context, in a different country and language. How the reader will respond to this offer of information will depend upon their own institutional values and protocols.

Further, House also gives us a theoretical insight into a feature which is very widespread in official document translation, notably in Australia and increasingly around the world: the use of extract or template translation. In her discussion of translations and different versions of translation, she uses the notion of versions to show how translations can be directed to different audiences (e.g., for a youth readership of a novel), or

when the ‘translation’ is given a special added purpose. Examples are interlingual versions or ‘linguistic translations’, résumés and abstracts, where it is the express purpose of the version producer to pass on only the most essential facts of the original. (House, 2014, p. 256)

This then provides us with an analogy of the extract or template translation functioning like an abstract for a larger piece of text. How either translation quality assessment or translation quality assurance stand in relation to this even more constrained version of translation is what we will explore in the empirical part of this paper.

When speaking of the function the translation performs, the quality criteria for official translations, as already indicated, must include one additional element beyond just comparison of source with target text and versions thereof – namely, the degree to which the users of the translation (regulatory, educational and official bodies) can accept it as clearly conveying the particular identity or status purportedly claimed in the source document.

Theoretically, the official document is not merely a description of an identity or status – the document itself announces a particular state of affairs and a claim to the recognition of this statement of identity or status. We can make an analogy with the long-established linguistic theory of performative utterances, where a particular speech act itself conveys a specific status – ‘I pronounce you man and wife’, ‘I swear to tell the truth’ (Austin, 1962). Official documents thus represent an announcement of a status (e.g., a driving licence allows the holder to drive on public roads), so they are the epitome of a performative document, granting of themselves a certain status to their bearer. The question of quality then relates not only to how well the translation mirrors the source text in linguistic terms, but how well it allows the reader of this translation (in another country and perhaps a quite different institutional context) to use the translation to grant or not grant an equivalent status in the receiver society.

A survey including both translators and institutional end users (Taibi & Ozolins, forthcoming a) shows the two groups to be very clear in seeing the translation task as not to prejudge how the bearer’s translated document will be assessed, but to reproduce as accurately as possible how the original institution describes the relevant identity, status or qualification. One professional regulator described the very exacting requirement their office placed on translations (full translations rather than extracts, statement of accuracy from the translator, extra precautions for translations done in other countries, checking of translator credentials), all to give the assessing officials the best chance of making a reliable qualification assessment. In response to a question as to whether there were complaints about the quality of translations, the regulator ironically replied:

Our applicants usually complain about the difficulties they encounter arranging for the documents to be translated in accordance with our guidelines, rather than the quality of the translations.

Translators were uniform in describing their own task as not presupposing equivalences in documents of Australian norms, but to offer as good a version as possible of the source document.

The English translation of those key [qualifications] should be direct, not free interpretation, but at the same time meaningful and not confusing or misleading. No attempt should be made to put what the translator may believe is the equivalent of overseas qualifications in Australia, which is the role of various institutions in Australia that assess qualifications gained overseas. This approach makes the job of those institutions easier and shows the client that it’s not the translator that assesses their qualifications, which is what many clients actually expect with some even suggesting what the translator should put as being, in their opinion, the “correct” translation, or simply one that they may believe might help their cause.

The end user is a constant hovering presence to the translator of official documents, and the views of many of our translator respondents are well encapsulated in the following contribution which, while made with the end user specifically in mind, also addresses the knowledge gap between readers and translators:

I would be thinking about whether the reader of these documents would be able to understand what I'm translating. It's not in terms of language, but in terms of the context, because as translators, we all have this knowledge about different cultures in different countries. But our readers don't necessarily have that. So it is my job, I think, to translate, so number one is to translate the document as accurate [sic] as possible. And the other thing, too, makes make [sic] sure that my translation product is understandable so the worst thing you would have is a reader having this, and think about what is that? […]

Our analysis of the corpus of translations in this paper will show how far these quality objectives are met.

Translation quality assurance in the translation of official documents

We now return to the issue of the other ‘branch’ of quality, namely Drugan’s (2013) concern with how translators or translator companies can guarantee quality to their clients before any translation takes place. In our work on community translation (Taibi & Ozolins, 2016) we set out a number of stages of a translation project which would ensure maximum TQA for community translation texts commissioned for translation companies. The six steps were:

• Work on the source text – i.e., many texts submitted for translation were usually not written with translation or minority readerships in mind, and may have many features which render them problematic for intended readerships (e.g., institutional assumptions, unexplained processes or requirements, or simply poor writing). Good translators will comment on these features to commissioners, requesting clarification or change.

• Selection of translators

• Briefing of translators

• Ensuring project management focus – that is, ensuring the translators become part of the project management processes

• Ensuring communication with author or commissioners

• Ensuring quality checking or other post-draft quality assurance features.

When considering translations of official documents, two of these steps are rendered impossible simply by the inherent constraints of these documents – the source text cannot be interrogated, clarified, or rewritten in any way; and communication with the original authors is in almost every case impossible. This is not to say that researching of source country institutions and processes would be of no avail – in fact it is often a necessity, as our survey of practitioners reveals – but the source texts remain immutable. Two other criteria are almost as difficult to meet: project management and translation checking. Ensuring a project management focus is difficult unless the translators are working on a continual body of such translations in relation to project management. As for post-draft checking, translations of official documents are not always subject to revision, a necessary process in all substantial translation (Mossop, 2019). Due to the generally concise nature of the documents involved – commonly a single page, running to a few pages at most – and the logistics and practicalities of engaging external revisers (short texts, privacy and non-disclosure agreements), it is almost universal practice for one translator alone to complete the translation.

Thus, we are peculiarly bereft of many of the quality assurance features that we can assume for other areas of translation (commercial, technical, legal, community etc.), which is testimony again to the particular nature and function of official translations, and their status as being outside the usual remit of community translation and more in line with very specific international communication functions.

However, this is not to say that quality issues are not addressed by either translation companies or practitioners, or that their own TQA measures are not significant.

Quality mechanisms at practitioner and company level

While formal revision is rarely commissioned, the fact that a good proportion of such translations will be commissioned by translation companies operating in the host society, and the translation will overwhelmingly be into the language of the host society, means that many translation company employees can view and comprehend these translations.

In no other field of translation is the quality aspect of the agencies and translation companies who outsource their translation perhaps as critical as in official document translation, for when dealing with translations into multiple other languages, there is little that company staff can do beyond formal requirements (format appropriately, check prima facie completeness). But in this case, in Australia, where official translations are preponderantly into English, any quality assurance mechanism the company does have comes into its own.

As reported in another paper (Taibi & Ozolins, forthcoming a), the surveyed Australian translation companies and agencies all strongly assert their responsibility to ensure quality when processing the translations coming back from their freelancers – from the survey of language service providers, we note that companies see their own integrity to be at stake:

Personal document translations undergo a stricter proofreading process because we believe that an error (misspellings and misinformation) in these types of documents can delay our clients’ processing of these documents on their end.

In particular, the LSPs check for content accuracy and idiomatic language use, as well as punctuation and formatting. This is seen as critical quality assurance, where companies find errors made by their contract freelancers, albeit often for understandable reasons of haste or inattention, including typos and oversights such as cutting and pasting document headings or holder details.

As one LSP commented from an essentially operational standpoint: ‘[q]uality is defined by the amount of “correction requests”’. In all companies’ experience, such correction requests are very rare, and almost always in relation to the rendering of names in Latin script.

But there is one other step that companies can take to ensure quality, as commented on by several LSPs:

After translation, we usually send a draft first to the client to confirm if all details (names, dates, addresses) are correct. Upon confirmation from the client, we then provide the final document. (Taibi & Ozolins, forthcoming a)

This relates to a significant feature of the status of clients: most of the individuals requesting these translations are fluent enough in English to be able to check the result; particularly in relation to the educational qualification cases which feature so prominently in our study, clients are often attempting to gain entry into work where high English competence is a prerequisite, and thus are well equipped to spot errors or comment on appropriateness of translation.

With these mechanisms, translation companies believe they have as robust a TQA regime as can be built, although this of course would not be the case for translation companies that see themselves only as intermediaries (‘letterboxes’) between translators and commissioners.

However, as our survey of practitioners reveals (Taibi & Ozolins, forthcoming b), a sizeable proportion of these translations do not come through translation companies but are commissioned directly by the holder of the documents. In these cases, although in our survey there are a few solitary instances where translators mention outsourcing of review tasks (usually cooperative and informal, as in seeking help from another practitioner), the translator is overwhelmingly on their own.

From a practitioner perspective, the constraints on producing quality translations stem overwhelmingly from the poor quality of the source documents (Taibi & Ozolins, forthcoming b), including omissions, fading, illegibility, errors, and inconsistencies within and across documents. In content terms, some challenges relate to certain institutions, educational practices or categorisations, even vehicle types, that are unknown or without direct equivalence in Australia, and thus require explication to help ensure end reader comprehension.

Practitioner understanding of quality requirements is very consistent across the survey sample: emphasis on accuracy and exact mirroring of the source is ubiquitous, as the following examples show:

1. Word for word, sentence for sentence are almost the rule of thumb in translating an official document. […] While we should be strict in accuracy because they are official documents, we do not undermine the fluency of the translated text.

2. The translation has to reflect the content and layout of the original document as much as possible. Authorities are very strict and there is not too much room for creativity.

3. A quality translation is an accurate, precise and transparent translation.

4. Attention to detail and accuracy are the key quality criteria in the translation of official documents.

In the following section we discuss the assessment method we applied to the body of official document translations under examination, and subsequently use this to illustrate a methodical (post-facto) quality assessment approach with a view to informing moves toward more widespread adoption of quality assurance among the essential processes of handling and translating official documents.

Research methods

This article reports on one phase in our broader project concerned with translation of official (personal) documents,¹ namely the assessment of 300 sample translations from three languages: Arabic, Chinese and Spanish. The aim of the assessment phase was to compare the translation approaches adopted in three different languages and identify areas of good practice as well as areas for improvement, especially in relation to translation quality and integrity.

A total of 300 copies of source documents and their respective translations (100 in each language) were collected from different translators who were all certified/accredited by the National Accreditation Authority for Translators and Interpreters (NAATI). The three languages chosen represent a significant portion of non-English speakers in Australia.

An initial email announcement provided general information about the project and advised potential participants that the researchers were seeking samples of translated personal documents (together with their originals in one of the three languages). Translators who expressed interest were informed that the documents of interest were typical personal documents such as qualifications, police clearances, civil registry certificates, and driver licences. They were asked to share scanned copies of de-identified documents and translations. To ensure diversity, they were also advised that the maximum number of documents per translator was 15 and that the documents should be of different types (e.g., 2 birth certificates, 3 driver licences, 2 educational certificates, 1 police check, etc.).

The source and target documents were de-identified to remove both sensitive holder details and translator information. To ensure confidentiality, some translators removed their stamp and details completely; others redacted their name and NAATI number only. This double confidentiality measure (applicable to both translator and holder details) had some impact on the information and features available for assessment. For example, redacting names and surnames meant that these details were no longer available to check in terms of accuracy. Similarly, the absence of translator stamps in some cases obscured some of the security features of the translation (i.e., translator certification).

The translations were subsequently assessed by two assessors in each language combination. The assessors were NAATI examiners and/or tertiary educators with professional and teaching experience in translation – and translation of official documents in particular. They received the scanned copies exactly as they were submitted by the professional translators, except where a source document or translation required further de-identification. They were informed that we needed to conduct an overall evaluation of the translations in question. This evaluation would identify the features of effective/appropriate translations and any trends in professional practice that might arise in different working languages. They were additionally informed that this qualitative analysis would also identify best practice that caters for the expectations of public services, language services and professional standards.

The assessors were asked to use a rubric developed by the researchers based on existing literature (e.g., House, 1977 and 2014; Nord, 1997/2018 above; also Mayoral Asensio, 2014, especially in relation to the translator as public authenticator), NAATI translation testing criteria and standards, and findings from the first stage of the current project (surveys with translators, language service managers, and end users). The rubric focuses on four main areas: accuracy, integrity, language and style, and presentation.

1. Accuracy:

   • Accurately translates propositional content and intent of source text

   • Accurately reproduces details such as names of persons, places and administrations

   • Includes all essential information for type of document

2. Integrity:

   • States whether the document sighted is an original, a certified copy or a non-certified copy

   • Describes official features of the source document (stamps, signatures, etc.)

   • Describes unusual features (obliterated text, cross-out, or corrections)

   • Uses square brackets for the translator's in-text descriptions or notes, or uses translator's notes elsewhere in the translation

   • Includes translator disclaimer, such as the one recommended by AUSIT (2014, p. 8) ‘The translator, in providing this certification, gives no warrant as to the authenticity of the source document. Any unauthorised change to the translation renders this certification invalid’.

3. Language and style:

   • Uses formal register appropriate for the translation of official documents

   • Uses language structures and lexicon correctly

4. Presentation:

   • Presents contents in a clear, organised and user-friendly manner

Each item of the above was allocated a column in a spreadsheet, and the assessors were instructed to fill in each column for each translation with ‘yes’, ‘no’ or ‘not applicable’. The assessors were advised that the researchers were aware that a binary system did not provide the best way to reflect the details of quality and integrity, and that in some columns it would be difficult to determine whether ‘yes’ or ‘no’ applied, but for ease of quantification and comparison they were asked to insert ‘yes’ where an affirmative answer was mostly the case for the criterion in question, or ‘no’ where the translation mostly failed to meet that criterion. For example, regarding ‘accurately reproduces details such as names of persons, places and administrations’, if the translator mostly did so then the assessors were expected to enter ‘yes’. If the translation mostly failed to reproduce those details accurately, then the assessors were instructed to enter ‘no’.

In addition to the above, the assessment rubric also included two columns for an overall evaluation:

• (5) Overall mark out of 100: taking into consideration all the quality, integrity and presentation aspects above;

• (6) Aspects that need improvement: dot points or brief comments on the areas that needed improvement in each translation.

For the summative assessment of the translations, a decision was made to consider an inter-rater difference of 10 marks or fewer as acceptable. This was based on two reasons: (1) When NAATI testing used a numerical assessment system, a third marker was called in when the difference between one examiner and another was greater than 10 points; and (2) In translation assessment, it is extremely unlikely that two examiners will agree on exactly the same numerical value out of 100. With a tolerance of up to a 10-point difference between raters, the inter-rater reliability for the translations from Chinese was 92% (92/100 = 0.92), followed by 76% for the translations from Arabic (76/99 = 0.76) and 63% for the Spanish group (61/96 = 0.63). Given that an adequate level of agreement is generally a minimum of 70% (Multon & Coleman, 2018), only the Spanish assessors show a level of agreement that is slightly lower than adequate, while the level is very good in the case of Arabic, and excellent for Chinese.

Findings

Judging by the average overall marks given by the two assessors in each language combination, the corpus of translations of official documents generally demonstrates a high quality standard. As Table 1 shows, the average mark was 92.83/100 for the translations from Chinese, 83.74/100 for the translations from Arabic, and 83.17/100 for those from Spanish.

Table 1. Average marks.

Language	Average mark Rater 1	Average mark Rater 2	Average mark both raters
Arabic	88.91	78.58	83.74
Chinese	93.62	92.05	92.83
Spanish	86.2	80.15	83.17

The overall figures (above 80/100) suggest a high level of satisfaction with the quality of most of the translations. Considering that until 2018 the numerical threshold for the NAATI Professional Translator credential (under the former accreditation system) was 70/100, the overall quality of the translations may be regarded as at or above a minimum professional standard. However, there are a few translations in the corpus that do not attain this quality threshold. Taking the assessment of the ‘harshest’ examiner in each language combination, 15 translations from Chinese and 15 from Spanish were rated below 70% (2 as low as 40% in both language groups), while 12 translations from Arabic were rated below 70% (a couple of them as low as 40%).

The areas for improvement identified in the Arabic assessors’ feedback were mainly related to:

• completeness of information (information missing in the translation)

• the need for the type of document sighted to be specified (original, certified copy, etc.)

• accuracy (distortion of meaning, incorrect lexical choices and terminology), and

• language (incorrect syntax and register, and minor spelling errors). An example of language errors is the following: ‘He finished his military service in the Airforce, with good record for a service for a period of 1 year, 1 month and 20 days’.

For the translations from Chinese, the main areas for improvement were:

• accuracy (minor and major distortions)

• completeness (minor omissions; extract translation was also mentioned several times)

• missing, incomplete or poorly expressed translator disclaimer (e.g., ‘I …  … hereby certify that this is a true and correct translation of the Chinese document’ or ‘This is a true and accurate translation of the text provided on the attached document’)

• language (register insufficiently formal for official documents, unidiomatic English), and

• description/translation of official features (e.g., stamp copied and pasted with no description, omission of photograph and stamp, missing essential description of document type).

In the case of the translations from Spanish, the first area for improvement was the need to specify the source language (translated/translation from Spanish) and the type/heading of the document (e.g., Marriage Certificate, Police Clearance). This was followed by:

• the description of official features (e.g., coat of arms or logo should be described between brackets, not copied and pasted), and

• the need to specify that the translation is an extract translation when that is the case and the need to translate every detail in the source document when it is a full translation.

Several instances of inaccuracies (distortions, omissions) and improvable usage were also identified among the Spanish-English translation sample (inappropriate lexical choices, unidiomatic language). Other areas identified include:

• false equivalence with the Australian system,

• inconsistent and unclear use of translator notes (e.g., in some cases the boundary is unclear between translation and translator note), and

• translation of Catalan text within the source document, when it is unclear whether the translator is certified for that language.

To have a more detailed view of the quality and integrity aspects of the translations, we conducted a statistical analysis of those translations in terms of compliance with the eleven assessment criteria mentioned above and set forth in Table 2 below.

Table 2. Overall assessment of translations by assessment criterion.

	Accuracy			Integrity					Language and style		Presentation
	Accuracy of content and intent	Accuracy of details (e.g., names of persons and places)	Includes all essential information for type of document	States whether original, certified copy or non-certified copy	Describes official features of source document	Describes unusual features	Uses [] for translator’s notes	Includes translator disclaimer	Uses appropriate register	Uses language structures and lexicon correctly	Presents contents in a clear, organised and user-friendly manner
Arabic	84.84%	92.42%	74.74%	67.67%	88.35%	11.11% (Mostly N/A)	30.8% (Mostly N/A)	69.19%	82.32%	72.72%	93.93%
Chinese	100%	99%	96.5%	49.5%	82.5%	42.5(Mostly N/A)	73.5%	74.5%	99.5%	98%	99.5%
Spanish	89.06%	89.06%	95.31%	5.20%	88.54%	1% (Mostly N/A)	85.93%	65.62%	99.47%	95.83%	93.75%

As Table 2 shows, the translations consistently score high in most areas and across the three languages. This includes, among other aspects, accuracy of propositional content (84.84% for Arabic, 100% for Chinese, and 89.06% for Spanish), description of official features in the source document (88.35% for Arabic, 82.5% for Chinese, and 88.54% for Spanish), correct use of grammatical structures and lexicon (72.72% for Arabic, 98% for Chinese, and 95.83% for Spanish) and use of appropriate register (82.32% for Arabic, 99.5% for Chinese, and 99.47% for Spanish).

The areas where scores are lower than expected are (1) document sighting, i.e,. specifying whether the translator had access to an original source document, a certified copy or a non-certified/electronic copy (for Spanish, the percentage was as low as 5.2%, for Chinese it was nearly 50%, while for Arabic it was slightly higher, 67.67%); and (2) translator’s disclaimer, which was missing in approximately 35% of the translations from Spanish, 30% of those from Arabic, and 25% from Chinese. The scores for the description of unusual document features (e.g., text erasures, redactions or corrections) were very low too, but that was due to the fact that in most cases no such features were present in the source documents.

Discussion

The overall picture that emerges from the assessment of our corpus of official document translations shows a high level of translator performance and translation, quality. This is consistent with the fact that the translators in question are duly certified/accredited by NAATI. However, the existence of a small percentage of sub-standard translations (15% from Chinese, 15.62% from Spanish, and 12.12% from Arabic) is also a concern. In addition, in an area such as the translation of personal official documents, even work that is done to an otherwise high standard may still contain errors or omissions in the transcription or annotation of holder details, numerals and other document particulars which, while perhaps ‘minor’ in terms of linguistic transfer, can present major inconvenience or risk for clients and institutions. As pointed out above, areas for improvement suggested by the assessors include completeness of information, document sighting (specifying the type of source document sighted), accuracy, language (both in terms of grammar and register), description of the official features of the source document, and inclusion of a translator’s disclaimer. Several translations which were rated at or above 80/100 also received feedback about minor inaccuracies, incorrect or inappropriate lexical choices, missing information, missing document features, and so on.

In another publication (Taibi & Ozolins, forthcoming b), we found that the translators working with this type of document were self-reportedly able to deal with the attendant challenges in most cases; however, a small group acknowledged unresolved difficulties and uncertainties, as well as limitations in professional knowledge and skills. Although it can hardly be claimed that the sample of translations assessed in this study are fully representative of the language combinations in question or the general professional practice in Australia, it does suggest that challenges are generally successfully managed. However, the exceptions call for further quality assurance measures. In Taibi and Ozolins (forthcoming b), some translators rightly noted that quality assurance is not only the translator’s responsibility, but also that of other stakeholders, particularly language service providers and translation agencies. In this regard, they called for more consistency in quality assurance approaches and procedures.

Another area where more consistency and clarity would help is the use of, and choice between, extract or full translations. Many of the instances where assessors noted that the translation was incomplete concerned extract translations, some of which did not specify the translation approach (full or extract). As pointed out above, template translation enjoys very widespread use for official documents in Australia, and increasingly in some other parts of the world. Theoretically, this is a legitimate option in line with House’s (2014, p. 256) notion of ‘version’ mentioned earlier, i.e., a translation product intended for a specific audience and aiming ‘to pass on only the most essential facts of the original’. From a practical and industry perspective, most Australian translators (53.04%) were happy to use extract translations because they are more efficient and economical than full translations (Taibi & Ozolins, forthcoming b). However, there were others who had concerns about the risk of leaving out relevant or important information. While most Australian institutions do accept template translations, professional accreditation and educational institutions are prominent in demanding full translations; there is no general rule as such, and not all end users clearly advertise their own policy in this regard. It follows that more institutional transparency would reduce uncertainty in this area. As for translators, a clear conclusion that emerges from the corpus assessment is that they must always indicate in their translations whether these are full or extract.

A related area for improvement is translator training in general, and training in the translation of official documents in particular. We reported elsewhere (Taibi & Ozolins, forthcoming b) that 64.34% of the surveyed NAATI-certified/accredited translators had never completed training or a professional development activity specifically covering this area of translation practice. While some of the areas for improvement identified in the current study (e.g., accuracy, language skills and register appropriateness) apply to all translation fields, there are field-specific shortcomings that can be addressed through relevant modules in translation programmes and/or periodic professional development workshops. Practicalities such as the description of official features (stamps, coats of arms, signatures, etc.), document sighting (original/certified copy or electronic copy), translation approach (full or extract), specifying the source language, translator disclaimer, and so on do not require much learning effort, but can make a difference in terms of translation, quality, integrity and reliability and, therefore, in terms of end user satisfaction and trust.

The translation of personal official documents is often perceived as ‘low level’ translation, but pedagogy needs to break this stereotype and stress the importance of this area of translation. Translator education programmes need to raise awareness of the critical nature of personal document translation as well as awareness of the vast range of documents this covers. The geographical, institutional and typological diversity of these documents presents challenges that should be addressed in training programmes.

Conclusion

The translators of official documents are public authenticators (Mayoral Asensio, 2014, pp. 3–4) in the sense that they translate for public entities and certify that the information conveyed in their translations is accurate and can therefore have the ensuing legal, health, educational or other implications. This translation practice is as essential to the interests of individuals moving between nations and administrative and legal systems as for institutions dealing with clients from different language backgrounds.

Given the real-life consequences of this area of translation practice, it is often strictly regulated (through educational qualifications, professional testing, or other forms of accreditation). In Australia, translators must be certified by the National Accreditation Authority for Translators and Interpreters (NAATI). However, translated official documents, because of their private nature, are generally inaccessible to people other than the translators, clients and public service staff processing them. This reduces the visibility of any quality or integrity issues that might be having an impact on the clients or end users. The literature on translation quality also shows little interest in the translation of official documents, which suggests that any improvements in this area would come from the industry itself, rather than from scholarly literature.

In this study, we managed to collect 300 samples of official translations (from Arabic, Chinese and Spanish into English) and assessed them with the assistance of independent examiners and educators. We created a tailored assessment rubric based on relevant literature (House, 1977 and 2014; Nord, 1997/2018; Mayoral Asensio, 2014), NAATI’s translation testing system, and the views of some relevant stakeholders (translators, language service managers, and end users). The rubric included three main categories which apply to other areas of translation as well (accuracy, language and style, presentation), but also, and most importantly, criteria relating to translation integrity and description of official features of the source document, which apply specifically to this type of translation.

The assessment of the translation corpus shows that the appropriate quality standard is met to a high degree, and reflects the performance expected of NAATI certified translators. At the same time, there was a small percentage of translations that were below the expected standard. The assessment also identified areas for improvement such as translation completeness or integrity, information accuracy, and language appropriateness and naturalness. Particular shortcomings include failure to specify the type of document received for translation, failure to specify whether the translation is a full or a template one, failure to include a translator statement (public authenticator’s disclaimer), and failure to describe or translate the official features of the source document. Thus identified, these quality and integrity issues can be remedied through appropriate education and training, and are of clear interest to all parties (researchers, educators, practitioners, LSPs) who are active in the field of official translation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Multicultural NSW and the National Accreditation Authority for Translators and Interpreters (NAATI).

Notes on contributors

Mustapha Taibi

Mustapha Taibi is Associate Professor of Interpreting and Translation at Western Sydney University, and Editor-in-Chief of Translation & Interpreting. He has conducted a number of studies on community interpreting and translation. His recent publications include Community Translation (co-authored with Uldis Ozolins, Bloomsbury, 2016), Translating for the Community (Multilingual Matters, 2018), Multicultural Health Translation, Interpreting and Communication (Routledge, 2019, co-edited with Meng Ji and Ineke Crezee), and Translating Cultures (Routledge, 2021, co-authored with David Katan).

Uldis Ozolins

Uldis Ozolins is Adjunct Associate Professor at Western Sydney University and is an international leading scholar in language policy and Translation and Interpreting whose publications have had a high impact. He has taught Translation and Interpreting at several Australian universities and participated in several funded research projects as well as T&I professional improvement and professional development projects across the field. He has been active in the international Critical Link conferences and publications on community interpreting, and also runs a consultancy advising on language services policy.

Notes

1 The project ‘Translation of official documents: Ensuring quality and enhancing security’ was approved by the Human Research Ethics Committee at Western Sydney University (reference number: H13582).