1.2.1 Linguistics

Linguistics is the scientific study of language in general, encompassing scientific study of specific languages (but not mere ability to speak them, nor study of their literature and culture as arts or humanities disciplines) and scientific study of interpreting and translation. Applied linguistics is the use of linguistic knowledge to assist in practical applications. Various branches of linguistics investigate structural, social, cultural, psychological and computational aspects of language and language use. In this Element we use the general term ‘linguist’ to refer to scholars in any branch of linguistic science, with more specific terms when relevant.

While there is wide variation in the content and orientation of the branches of linguistics, all are based on comprehensive and systematic analysis of empirical evidence, using scientific methods, including observation, hypothesis and analysis.

These methods have enabled linguists to establish an impressive array of scientific findings on a range of topics.

Importantly, many of these findings directly contradict apparent ‘facts’ about language that are accepted by many people as part of our society’s body of ‘common knowledge’. A frequent theme in linguistics is that common knowledge about language embodies some misconceptions, sometimes called myths. This creates problems especially for branches of applied linguistics which involve direct interaction between linguists and non-linguists.

1.2.2 Forensic Linguistics

Forensic linguistics is a branch of applied linguistics that uses expertise in linguistics for applications related to legal contexts. Though the origins of the discipline can be traced considerably earlier, the first use of the term ‘forensic linguistics’ is usually ascribed to Reference SvartvikSvartvik’s (1968) account of his linguistic work on a case in England in which the written record of a police interview included an apparent confession. Svartvik demonstrated that the confession had been fabricated by police – a practice that later came to be known in Australia and the UK as ‘verballing’ (falsely stating that a suspect had made a verbal admission). Verballing continued as an increasingly serious problem until the introduction of mandatory electronic recording of police interviews with suspects (but see Section 3.1.2).

Since then, forensic linguistics, though still relatively young, has grown to be a well-established interdisciplinary field, with numerous researchers and practitioners around the globe. It has a large literature, several international associations, including, among others, the International Association for Forensic and Legal Linguistics (IAFLL), the International Association for Forensic Phonetics and Acoustics (IAFPA) and a number of journals, including International Journal of Speech Language and the Law (IJSLL) and Language and Law/Linguagem e Direito (LLLD).

Forensic linguistics can be divided into two interrelated branches, sometimes called ‘legal linguistics’ and ‘forensic linguistics’ (this means ‘forensic linguistics’ is sometimes used as a general term and sometimes more narrowly). Legal linguistics studies how language is used through all stages of the legal process, including court hearings, police and lawyer interviews and many more. While this work has various motivations, one notable theme is a concern to ensure inclusive justice for all, especially for disadvantaged groups, such as Aboriginal people, or those who lack proficiency in spoken or written English.

Forensic linguistics (narrow sense) develops methods for analysing and evaluating language to be used as evidence in courtroom hearings, such as written documents or recorded conversations that provide information relevant to deciding the verdict. Again, this work has multiple motivations, but a notable theme is a concern to ensure justice in the use of language as forensic evidence, despite the widespread misconceptions about language that are embodied in common knowledge.

The present publication aims to give an account of the origins, development and current state of forensic linguistics in Australia from its beginnings in the late 1950s to the present day. While our field is interdisciplinary, we have necessarily restricted the scope of this Element to the work in legal contexts of those trained and qualified in linguistics. A wider focus is beyond the scope of this publication. However, work on language and law by lawyers and interpreters is of great interest and importance to forensic linguists (see Reference Grey and Smith-KhanGrey & Smith-Khan, 2021).

1.3.1 Precursors 1959–1970s

Linguistic evidence was first presented in an Australian legal matter in 1959, almost a decade before Svartvik coined the term ‘forensic linguistics’. The evidence was given by then Reader (later Professor) in Linguistics at Adelaide University, T. G. H. Strehlow, in the South Australian case against Aboriginal man Rupert Max Stuart. Stuart’s murder conviction had relied on a confession that Strehlow argued was fabricated, that is that Stuart had been verballed.

Strehlow’s expert evidence revealed insights about language variation that were ahead of the field of linguistics in the late 1950s, describing Stuart’s way of talking English initially as Pidgin English, and then Northern Territory English, that is, ‘the language spoken by most educated part-aboriginals’ (note that the term part-aboriginals is no longer acceptable). A decade later, the ground-breaking work of American linguist William Labov on (American) Black English Vernacular resonated with Strehlow’s (unpublished) description of Northern Territory English. Central to Labov’s work, which revolutionised modern linguistics, was the recognition that the way African Americans speak English is similar to a regional dialect, but it is a dialect whose speakers share ethnicity rather than geography. Although Strehlow used the seemingly geographical term ‘Northern Territory English’, he defined the variety in terms of the ethnicity of its speakers (see Reference EadesEades, 2013).

The late Gloria Brennan, a Western Australian woman whose first language was Wongi, was the first Aboriginal person to graduate in linguistics (1978). Her fifty-five-page report (Reference BrennanBrennan, 1979) on the need for interpreting and translation services for Aboriginal people was the first linguistic research to deal with language in legal contexts in Australia (commissioned and published by the Australian Government).

With the enactment of land rights for Aboriginal people in the Northern Territory in 1976, many linguists doing research in that jurisdiction became actively involved in the preparation of evidence for land claims. While their work on documenting (often complex) relationships between social groups and language and land has not generally been considered forensic linguistics, the work of Michael Walsh, Harold Koch and a few other linguists on linguistic aspects of this particular legal process is an exception (Sections 3.6.3 and 5).

1.3.2 Early Days (1970s–1980s)

It seems the first branch of linguistics to be commissioned to examine linguistic evidence for its value in identifying the perpetrator of a crime was phonetics. In the infamous Qantas bomb hoax of 1971, telephone messages were recorded in which the caller claimed a plane had been planted with a bomb that would detonate unless a large ransom was paid. Police asked Alex Jones, of Sydney University English Department, for assistance in determining the speaker’s accent. The story is taken up in Section 3.2.

From the late 1970s linguists across several additional branches of linguistics were involved in sporadic casework, as documented in Reference EadesEades (1994). Their work addressed several different legal questions, including speaker identification, degree of pronunciation similarity between two trademarks, disputed transcription of tape-recorded utterances, alleged fabrication of confessional evidence (verballing), authorship attribution, meaning of specific words in Torres Strait Creole and Aboriginal English, and cross-cultural communication issues impacting interviews with an Aboriginal person. However, except for Reference HammarströmHammarström (1987), this work was not documented in academic publications, and there was little sense at this stage of forensic linguistics as a unified branch of linguistic science.

Also during the 1980s, Diana Eades was becoming known for her research on the linguistics of Aboriginal English and Aboriginal communication styles. Through this work, she was contacted by defence lawyers concerned about the possible verballing of an Aboriginal client, Kelvin Condren. This resulted in one of her earliest expert linguistic reports, which showed that Condren was highly unlikely to have used language attributed to him in the so-called verbatim transcript of the interview (R v. Condren, 1987). Eades’ report was excluded from the appeal, and Condren remained in prison for several more years, until a journalist’s incontrovertible evidence showed he had been in prison at the time of the murder of which he was accused (Section 4.2.1; Reference EadesEades, 2013).

1.3.3 Defining a Discipline (1990s)

Experience in Condren and other cases led Eades to recognise a number of misconceptions and problematic attitudes about language that are embodied in legal practice. For example, lawyers and judges were reluctant to accept Aboriginal English as a rule-governed dialect of English. They saw language as related to biology (e.g. skin colour) rather than socialisation, and they believed linguistic judgments could be made on the basis of common knowledge without need for specialised expertise.

Eades developed a new research programme focused on documenting and countering problematic attitudes and presuppositions in the law, especially as they affected justice for Aboriginal people in the legal process. To bring attention of other linguists to these issues, Eades convened a popular annual workshop on language and the law at the conference of the Australian Linguistics Society (1989–1996). She also collated information about expert linguistic evidence in Australian courts, as well as a growing research base on forensic linguistics, to make it accessible to colleagues (Reference EadesEades, 1994, Reference Eades1995).

During these years, Eades became part of a group of linguists around the globe who established forensic linguistics as an international discipline through the International Association of Forensic Linguists (established 1992, renamed the International Association of Forensic and Legal Linguistics in 2021) and the journal Forensic Linguistics (established in 1994, renamed International Journal of Speech Language and the Law in 2003). This journal has always been shared with the International Association of Forensic Phonetics (established 1991, later renamed the International Association of Forensic Phonetics and Acoustics).

Eades also convened the 1995 meeting of IAFL at the University of New England Armidale, further catalysing interest in forensic applications among Australian linguists – some of whose work is described in this Element.

At Monash University, Heydon began working on research addressing the linguistic dimensions of police interviews with children, a field previously dominated by psychologists (Reference HeydonHeydon, 1998, Reference Heydon2005: 148–64). Heydon’s work contributed to the development of forensic linguistics in Australia as a complement to psychological research into child cognition and legal interviewing.

Meanwhile, also during the mid-1990s, Australian forensic phonetics was being established on a scientific footing through the work of Phil Rose, Andy Butcher, Helen Fraser, John Ingram and others in the Australian Speech Science and Technology Association (ASSTA), as recounted in Section 3.2.

1.3.4 The Twenty-First Century

By the late 1990s, forensic linguistics and forensic phonetics were well established in Australia, with expanding research across various branches (documented in later sections) and increasing interaction with international colleagues.

A number of students were undertaking postgraduate research, and undergraduate courses in language and law were started at the University of New England and the Australian National University, followed by others, including Macquarie University. While these courses are generally taught within linguistics departments, at Monash University Law School a course entitled ‘Language, Communication and the Legal Process’ has been developed by forensic linguists Peter Gray and John Gibbons and taught several times in the last decade (Section 6.2). Starting in the 1990s, courses in legal interpreting have been taught at Western Sydney University (formerly the University of Western Sydney), the University of New South Wales and RMIT University, among others.

There is now a new generation of forensic linguists, whose early career research is already being published, including Alex Bowen (Sections 3.6.3, 5.1 and 5.2), Alexandra Grey (Section 3.7), Ben Grimes (Sections 3.5.2 and 6.2), Elyse Methven (Section 3.7), Laura Smith-Khan (Section 4.2.4), Yuko Kinoshita (Sections 3.1.2 and 3.2.5), Shunichi Ishihara (Sections 3.2.4 and 3.4) and Miranda Lai (Sections 3.6.2 and 4.1.1). The first five of these came to forensic linguistics with legal training and experience (see Section 2.1).

The most recent development is the establishment in 2020 of the Research Hub for Language in Forensic Evidence at the University of Melbourne (Section 3.1.4).

1.4.1 Beyond Merely ‘Applying’ Linguistic Knowledge

Not all the questions for which forensic linguists use their expertise can be fully answered simply by ‘applying’ well-established research findings from linguistic theory. Throughout this Element, readers will see that a key feature of both branches of forensic linguistics in Australia (as in some other countries) is the way that linguists’ work both draws on existing research and leads to new research. Indeed, Australian linguists, like some others, have used forensic linguistics to enrich linguistic science more generally.

One way this can happen is via case studies which provide opportunities for analysis of primary data that might not otherwise be available to the linguistic researchers. Eades’ work on the Condren case, for example, resulted in clear evidence of the importance of including pragmatic features in the definition of Aboriginal English (Reference EadesEades, 1993).

In addition, some questions on which linguists are asked to provide expert evidence in court require the discipline not just to apply existing knowledge but to generate new knowledge. For example, while linguists have long-standing knowledge about the pronunciation differences that characterise regional accents of various languages, there is far less work looking at the extent to which those characteristics enable linguists and others to accurately identify speakers from particular regions (Section 3.3). The latter is an issue that came to prominence with the implementation of language ‘tests’ for asylum seekers from the 1990s (Reference FraserFraser, 2009).

Readers will note that the concept of context is relevant to almost every section of this Element. This reflects the central role of context in many branches of linguistics. While we can follow leading sociolinguist Jan Reference BlommaertBlommaert (2005: 251) in defining context broadly as ‘the totality of conditions’ in which language is being ‘produced, circulated and interpreted’, different branches pick out different aspects of context as a focus. An important consideration in analysis of language used as forensic evidence is that context also ‘primes’ the analyst’s interpretation of the evidence. While reliable context primes analysts in helpful ways that are necessary to reliable interpretation, it is essential to guard against exposure to potentially misleading context (Section 3).

1.4.2 Direct Engagement between Linguists and Judicial Officers

A prominent feature of forensic linguistics in Australia is extensive direct engagement between linguists and the judiciary (Section 6). This started with workshops and publications seeking to inform lawyers and judges regarding findings of linguistics that demonstrate the problematic nature of some of the attitudes and beliefs embodied in the law as ‘linguistic ideologies’ (discussed in more detail below).

Diana Eades and Sandra Hale have been especially prominent in this area. As discussed in Section 6, engagement like this has generally been well received in Australia. For example, both Eades and Hale have worked directly with the law to recommend legal practices and procedures for improving opportunities for equality before the law, for example, Reference EadesEades (1992), and Hale in the Recommended National Standards for Working with Interpreters in Courts and Tribunals (JCCD, 2022, hereafter National Standards). (See Sections 2.3.3 and 4.2.1.)

Later their work became an inspiration for the ‘Call to Action’, in which all four national linguistics bodies called on the judiciary to review and reform the handling of indistinct covert recordings used as evidence in criminal trials, as recounted in Section 3. This eventually led to creation of the Research Hub for Language in Forensic Evidence at the University of Melbourne (Section 3.1.4). In October 2022, a workshop of judges, lawyers and linguists convened by the Hub, in collaboration with Deakin Law School, resolved to pursue appropriate reform of legal procedures regarding covert recordings.

1.4.3 Research on Misconceptions and Ideologies

Concern about the effect of misconceptions and problematic language ideologies has grown beyond documenting and addressing individual problems. In Australia, as elsewhere, some linguists (e.g. Reference EadesEades, 2012; Reference FraserFraser, 2018; Reference Smith-KhanSmith-Khan, 2022) have chosen not just to address them but to make them a topic of study in their own right.

These linguists have undertaken research directly focusing on the misconceptions and ideologies themselves, documenting exactly how they can contribute to injustice and unfairness before the law and recommending appropriate changes that can be made. As analysing and exploring the nature of linguistic misconceptions and problematic language ideologies has become such a prominent feature of Australian forensic linguistics, we devote a separate section to it, next.

2.2.1 Pre-Existing ‘Common Knowledge’

With only rare exceptions, all human beings can use spoken (or sometimes signed) language with great skill. It is often assumed that this skill implies underlying knowledge. However, there can be a disjunction between skill in using spoken and written language, and knowledge about how language works. This is similar to the famous observation that one can ride a bicycle without understanding the physics of how the bicycle works. However, the situation with language is worse than just lack of knowledge.

In western society, nearly everyone learns a basic stock of ‘facts’ about language through school education, especially literacy acquisition. For example:

• speech is a sequence of words, each composed of a sequence of ‘sounds’ which can be represented with written symbols

• words have fixed meanings, which can be looked up in a dictionary

• different languages represent the same meanings but with different words

• good grammar follows rules of logic

• effective communication results from sending a clear message

• people from different places speak in definably different ways

The problem is that some of the information about language imparted by school education has limited validity. For example, while the dot points above are accepted by many as basic facts about speech, all are recognised in linguistics as being at best partially true. Some are considered to have the status of myths, researched in specialised branches sometimes called ‘folk linguistics’. Misconceptions like these can be very hard to overcome. As teachers of linguistics at university level know, it can take years of study to recognise their fallacy.

The existence of such misconceptions does not affect the everyday skills of using language, but it can cause problems when the inaccurate knowledge is used for practical applications.

2.2.2 Lawyers’ Knowledge about Language vs Linguists’ Expertise

Higher education, especially in the professions, brings increased skill with spoken and written language, and increased knowledge about language, e.g. advanced spelling, grammar, etymology, knowledge of additional languages. However, unless it involves advanced study in linguistics, higher education typically deepens and elaborates the basic misconceptions of common knowledge rather than overturning them in line with the research findings of linguistic science.

Thus lawyers often have high level skills in the use of spoken and written language, and a strong sense of confidence in their knowledge about language. This is valuable in their profession, which depends heavily on language. However, it is different from the scientific orientation of linguists, who treat language as an object of study.

Indeed, in the early days of forensic linguistics many lawyers did not really know what linguistics was – thinking of a linguist as a polyglot, or as a grammarian enforcing proper usage (not realising linguistics takes a descriptive, rather than a prescriptive, approach). Now, however, there is far more recognition in the law of linguistics as a science – though this may involve a sense that it merely adds technical detail to common knowledge, without full recognition of the extent to which scientific findings can undermine apparent facts of common knowledge. The effect is that lawyers can sometimes feel free to override the opinion of an expert linguist. Some examples are discussed in Section 3.

2.3.1 Definitions

Widely held views about language are referred to in sociolinguistics and linguistic anthropology as language ideologies, which can be defined as ‘meta-level ideas about language’ (Reference Ford and MertzFord & Mertz, 2016: 10) or taken-for-granted assumptions about how language works that are socially, culturally and historically conditioned (following Reference BlommaertBlommaert, 2005: 253).

Language ideologies are typically accepted as basic common sense, but, like the common knowledge discussed above, they can be at odds with the findings of linguistic research. They are often accompanied by, or give rise to, practices (or ways of doing things) which are considered normal within particular social groups, in the sense that they are accepted without need for explanation or justification.

As mentioned above, Australian forensic linguists have studied a number of language ideologies in the legal process, many of which are not limited to the law. The focus of these studies is on how specific practices enabled by these assumptions about language can contribute to, or lead to, unequal or unfair treatment within the law of some individuals or some members of certain groups. This can create tension between the law and other institutions, and within the law, as social values evolve to embrace improved standards of fairness and inclusiveness.

The main language ideologies discussed in this Element are introduced below, as a frame for the coming discussion (see also Sections 4.2.3 and 4.2.4). This is just one way to enumerate and characterise language ideologies; others might classify them differently. Indeed, some scholars might discuss some of the misconceptions we have noted above within a language ideology framework. Here, however, we prefer to maintain the distinction between misconceptions and ideologies.

2.3.2 The Language Expertise Ideology

The language expertise ideology is related to, but different from, the issue of lawyers’ incomplete knowledge about linguistics discussed in Section 2.2.2. The language expertise ideology involves the assumption that skill in using a language necessarily implies the language user has explicit and reliable knowledge about the structure and usage of the language.

In relation to English language, this ideology underlies the legal expectation that juries can make their own evaluation of language evidence, without need for expert evidence. Examples are given in Section 3.1 and 3.2 of legal procedures that encourage juries to reach their own opinions, even if their view contradicts the opinion of an expert.

In relation to languages other than English, this ideology means the status of expert can be given simply on the basis of ability to speak the language. For example, witnesses can be asked to give an expert opinion purely on the basis of being a native speaker of the language (Sections 3.2.5 and 3.3).

2.3.3 Monolingual Language Ideology

A language ideology in many English-speaking countries, including Australia, is the assumption that monolingualism is, and should be, the norm in society. This is despite evidence that multilingualism is more prevalent in societies around the world than monolingualism.

This monolingual language ideology creates a range of problems, some stemming from poor understanding of what is involved in learning a language, or in living in a country where one has lower proficiency in the standard language. One effect noted by Australian forensic linguists is lack of sensitivity in the law to the difference between ability to produce a few words of English in answer to simple questions, and ability to participate fully in English-based legal proceedings.

As Section 3.6 will show, a number of Australian linguists have worked to counter this assumption. In particular, a correction to it underlies the results of Sandra Hale’s linguistic work on the JCCD (2022) National Standards (Section 1.4.2). Thus, part of Standard 16 for Judicial Officers advises them not to rely on ‘the … ability [of a person who speaks English as a second language] to provide biographical data as the basis for deciding whether to work with an interpreter’. This is because

Over recent years the legal system has made considerable progress in recognising that many Australians, both immigrant and Indigenous, need interpreters in order to understand and participate in the law. The National Standards document is arguably the clearest evidence of this progress. However, this is only a partial antidote to the monolingual language ideology. The assumption that monolingualism should be the norm is still widespread, and it is seen in a number of ways in the legal system, to be discussed in Section 3.6.

2.3.4 Written Language Ideology

Many ideologies that affect forensic linguistics can be seen as part of a larger societal ideology, with effects far beyond the law, that privileges written language as the basic form of linguistic expression (Reference OlsonOlson, 1994).

Although the law is often conducted in spoken language, it places a strong emphasis on written language, which is considered more stable and more ‘real’ than ephemeral spoken language. This is seen explicitly in the importance placed on written contracts and other legal instruments. However, it is implicitly involved in the priority given to transcripts of spoken language, including court proceedings, police interviews and many other events in the legal process (Reference Deamer, Fraser and HaworthDeamer et al., 2022).

The law has a rather unsophisticated view of the relationship between the transcript and the spoken language it represents, which are often seen as transposable analogues of one another (Reference HaworthHaworth, 2018; Reference FraserFraser, 2022a). In legal contexts this written language ideology can thus give rise to what Reference WalshWalsh (1995) calls ‘the tyranny of the transcript’.

The priority given to written language also has the effect, again seen in many areas within and beyond the law, that spoken communication is seen as message transfer, almost like sending a written note. This idea has been well studied as the ‘conduit metaphor’ (see Reference Lakoff and JohnsonLakoff & Johnson, 1980). The conduit metaphor creates the impression that the message is fixed, and simply travels from sender to receiver. This has been shown to be unrealistic, even for written language but especially for spoken language. Communication really involves the receiver making meaning of the message, which may align with the sender’s meaning to a greater or lesser extent (Reference Liddicoat and HaughLiddicoat & Haugh, 2009).

The conduit metaphor has been extended into the area of interpreting and translation, where it is useful in understanding a range of misconceptions about what is involved in rendering one language into another. For example, the conduit metaphor suggests, wrongly, that interpreters and translators simply exchange words and sentences from one language into another (see Reference HaleHale, 2004; Reference NakaneNakane, 2014; Reference FraserFraser 2022a). This has a major negative effect in the legal process, especially when combined with the monolingual ideology (Sections 3.6 and 5.2).

2.3.5 Referential Language Ideology

The referential ideology of language sees language as a fixed and transparent means of reference, as being ‘stable, denotational and context-free’ (Reference EhrlichEhrlich, 2012: 58; Reference EadesEades 2012; Reference Eades2016a; Reference Ehrlich and EadesEhrlich & Eades 2016). In this view of language, one of the simplest words in the English language, ‘yes’, when given in answer to a question seeking confirmation of a proposition, must always signal agreement to the proposition – because the meaning of ‘yes’ is taken to be stable, signalling agreement, and no contextual factors need to be considered in understanding its use in a particular answer. Despite the widespread operation of this referential language ideology in the law, we do not claim it is universal (for some exceptions see examples from judgments in Sections 3.5.1 and 3.6.1).

The view of language as having fixed meaning is aligned with the assumption (related to the conduit metaphor) that the primary function of language is to transmit information. This ignores the wide range of functions that language is used for, and the ways in which the meaning of an utterance can change depending on context (consider, for example, requesting, commanding, being sarcastic or joking).

To conclude Section 2, linguistic research in legal settings and linguistic expert evidence in specific cases highlight the central role of context in the workings of language.

For this reason, we have structured our report of the work of Australian forensic linguists in the next three sections in terms of linguistic, interactional and sociocultural contexts of the language data being analysed. We recognise that distinction of the three contexts is not always clear-cut and there is unavoidable overlap between the sections. This categorisation should be seen as one way of organising the material in a necessarily linear publication, but not the only way.

3.1.1 Legal History of Forensic Transcription

During the late 1970s, it became legal (under strict regulation) for police investigating serious crimes to use hidden listening devices to obtain covert (secret) recordings of suspects’ conversations. While these recordings were crucial during investigations, it soon became clear that they also had value as evidence in court. The first case in Australasia, and possibly in the world, where the prosecution sought to have this kind of surveillance audio admitted as evidence was a 1981 drugs trial in New Zealand (see Reference FraserFraser, 2021 for details on the case and the following history).

The problem was that the audio was of such poor quality that the jury could not understand what was said, even after hearing it played through twice. The prosecution offered a transcript prepared by police. Initially, the judge refused to admit this as it conflicted with multiple legal principles. As one example, the law considers a transcript to be an opinion as to the content of the audio – and police are generally not allowed to provide opinions in court, only factual evidence.

However, when the judge tried listening to the audio with the transcript, he found it so helpful that he agreed it should be provided to assist the jury in understanding the content of the recording, which, if they accepted the transcript, was relevant and probative evidence. This required adjusting the legal principles he had cited earlier. For example, since opinions can be given by experts, police transcribers were deemed to have a kind of ad hoc expertise, on the grounds that their ability to understand indistinct audio came from listening many times. These adjustments were upheld on appeal (R v Menzies 1982), and the case became a precedent allowing police transcripts to be provided to assist juries’ perception of indistinct covert recordings in many subsequent trials.

It is important for linguists to understand that police transcripts are not given directly to the jury. The law is well aware that a detective’s hearing might not always be accurate, and has developed a series of safeguards to ensure juries are not misled by an inaccurate transcript. Defence lawyers are expected to review transcripts and negotiate with the prosecution to create an agreed version for the jury. If agreement cannot be reached, the judge listens personally to ensure nothing potentially misleading goes to the jury. Finally, the judge instructs the jury that the evidence is the audio not the transcript, so they should listen carefully and reach their own conclusion about the audio content, using the transcript only as assistance (the ‘aide memoire’ instruction).

In 1987, these procedures were endorsed by the High Court of Australia (Butera v DPP (VIC) 1987), and eventually accommodated by the Uniform Evidence Acts that most Australian jurisdictions passed from 1995. Over the following decades, the use of covert surveillance spread rapidly, and the procedures became fully embedded across many branches of law and law enforcement. From a legal perspective, they are now completely standard, used routinely not just for covert recordings but for other types of forensic audio. However, from the perspective of linguistic science, they are problematic.

The procedures rest on the assumption that understanding spoken language is a matter of common knowledge (Section 2.2.1). This is why determining the content of audio evidence is considered a matter for juries, while development of safeguards to ensure juries are not exposed to potentially misleading transcripts is undertaken by judges. The problem is that, while understanding spoken language is certainly a common skill, knowledge about how that skill works, especially about the factors that affect perception of indistinct recorded speech, requires highly specialised scientific expertise (Reference Fraser and LoakesFraser & Loakes, 2020). Without that expertise, the law cannot effectively prevent listeners, including juries, from gaining confident, but inaccurate, perception of the audio, with potential impact on the verdict. In particular, the safeguards developed through legal precedent are not nearly as trustworthy as they seem to lawyers.

3.1.2 Involvement of Linguistic Science

Since the 1990s, the advice of experts has increasingly been sought when prosecution and defence cannot reach agreement on some part of a transcript (see Reference FraserFrench & Fraser, 2018), and evidence on disputed utterances has been given by Felicity Cox, Helen Fraser, Phil Rose and Andy Butcher, among others.

Through this kind of casework, Fraser developed concerns about the risk to justice posed by use of police transcripts, and especially about the assumption that the aide memoire instruction could ensure juries were not misled if the transcript happened to be inaccurate (see Reference FraserFraser, 2020b for more on this history). She led a 2011 experiment using audio from a real murder trial, demonstrating how easily an inaccurate transcript can mislead listeners into ‘hearing’ incriminating words which had not actually been spoken (see Reference Fraser and KinoshitaFraser & Kinoshita, 2021). This experiment attracted interest and concern from linguists – but lawyers were not disturbed. They trusted the safeguards, especially the concept that if the transcript was wrong, the defence would dispute it, and the judge would check it personally against the audio. The problem here is that even responsible listeners like judges can be misled by an inaccurate transcript.

This was forcefully demonstrated by another murder trial that came to Fraser in 2011. She showed that a police transcript that had passed all the legal safeguards, and been provided to ‘assist’ the jury, was actually misleading. In collaboration with psychologist Bruce Stevenson, she then used the audio in an experiment, showing that the lawyers and judges who checked the inaccurate police transcript must themselves have been unwittingly misled by it. This cast doubt, not only on the justice of the conviction, but also on the value of the safeguards (Reference FraserFraser, 2018). Reference FraserFraser (2013) argued that the situation effectively continued the possibility of verballing, supposed to have been eradicated by the introduction of recorded police interviews (Section 1.2.2).

Though it was too late for an appeal in this case, Fraser’s results were used for an application to review the conviction. The application was rejected on the grounds that legal procedures had been followed properly, so the jury were able to make up their own minds about the evidence (Reference FraserFraser, 2018). This marked a turning point in Fraser’s understanding of the causes of inaccurate transcription of indistinct covert recordings. It was now clear that the key problem with the legal handling of indistinct covert recordings was not the fact that transcripts are produced by police (though that is far from ideal). It was the fact that all transcripts, by police or experts, are evaluated by lawyers and judges, on the assumption that detecting and correcting errors requires only common knowledge. Fraser sought opportunities to explain scientific facts about perception of indistinct audio to the law. Through this work, additional problems with covert recordings were also uncovered. Particular risks were noted with lax procedures for admitting translations of indistinct audio featuring languages other than English. Other concerns included inaccuracy in attributing utterances represented in a transcript to particular speakers, and admission of ‘enhanced’ versions of indistinct audio.

3.1.3 The ‘Call to Action’

Fraser sought advice from the Australian Linguistics Society (ALS), who recommended a ‘Call to Action’. This was a 2017 letter, endorsed by four national organisations (ALS, ALAA, ASSTA and AusIT), asking the Australian judiciary to review and reform legal procedures for admitting indistinct covert recordings as evidence in criminal trials. The letter was delivered to Professor Greg Reinhardt, then Executive Director of the Australasian Institute of Judicial Administration (AIJA), who forwarded it to the Council of Chief Justices (CCJ). The CCJ assigned the ‘Call’ to Chief Justice Chris Kourakis, Chair of the Judicial Council on Cultural Diversity (JCCD), the body which had provided strong leadership on linguistic matters in the past (Section 2.3.3). In late 2019 a working group of four JCCD judges (Section 6.2) held an extended consultation with linguists Helen Fraser, Diana Eades, Georgina Heydon, Alex Bowen, Lesley Stirling (then President of ALS) and Peter Gray (see Section 6.2). Representatives from police and prosecution departments around the country were also invited.

At the end of the consultation, the JCCD judges acknowledged that the problems outlined by the linguists needed investigation, and in early 2020, the University of Melbourne’s Faculty of Arts established the Research Hub for Language in Forensic Evidence in their School of Languages and Linguistics, with Helen Fraser as Director and Debbie Loakes as Research Fellow (Reference FraserFraser, 2020a).

3.1.4 The Research Hub for Language in Forensic Evidence

In relation to forensic transcription, the Hub has two main aims. The first is to assist in reforming legal procedures to ensure that all indistinct audio admitted as evidence is accompanied by a demonstrably reliable transcript right from the start of the trial process. The second is to develop accountable evidence-based methods for creating reliable transcripts of indistinct forensic audio – for both English and other languages. This requires more than just applying existing scientific knowledge (Section 1.4.1). Although transcripts are used extensively in many branches of linguistic science, linguists rarely transcribe indistinct audio in situations where they do not know the content, and where the context is potentially misleading – which is the situation faced in forensic transcription (Reference Fraser, Bernardasci, Dipino and GarassinoFraser, 2022b). To ensure reliability, it is necessary to pursue cross-disciplinary research questions focused on demonstrating the capabilities of transcribers, not just on analysing the acoustic-phonetic properties of speech (see Reference FraserFraser 2020a, Reference Fraser, Bernardasci, Dipino and Garassino2022b for detailed discussion).

The problem is that no matter how good such evidence-based methods become, they cannot, on their own, ensure that juries are only assisted by reliable transcripts – since current legal procedures prioritise police transcripts even where they have been shown to be inaccurate by an expert on their own side (Reference FraserFraser, 2021). Law reform, though challenging, is essential.

In developing these ideas, it has become clear that, in addition to the situations described above, transcripts are used in concerning ways across other legal contexts (cf. Reference HaworthHaworth, 2018; Reference Deamer, Fraser and HaworthDeamer et al., 2022). This provides more evidence of the negative influence of misconceptions arising from the written language ideology (Sections 2.2, 2.3.4).

3.2.1 From Ear Witnesses to Forensic Audio

One way spoken language can be used as forensic evidence is to identify a speaker involved in a crime when visual recognition is not possible (e.g. if perpetrators wear masks). Traditionally, such evidence could only be given by a listener who had heard the voice live – an ‘ear witness’. Ear witnesses have long been allowed to testify in court, and their evidence is highly compelling. However, it is now known to be prone to error, with memory for voices even more unreliable than the poor visual memory of eyewitnesses, responsible for many miscarriages of justice (for more details see Fraser, 2019). Nevertheless, ear witness evidence is still allowed in courts around the world, including Australia (Reference McGorrery and McMahonMcGorrery & McMahon, 2016).

The rise of recording technology meant that in some cases voices were preserved (e.g. on a telephone answering machine). This reduced reliance on memory and extended the role of ear witness to listeners who were not present at the time. It also meant that juries, rather than relying on witness testimony, could now listen ‘with their own ears’, comparing the voice on the tape to that of the alleged speaker.

Further, they could be assisted in their judgment by testimony from witnesses familiar with the alleged speaker’s voice. Such witnesses included police reviewing surveillance audio – and the status of ‘ad hoc expert’ (Section 3.1.1) was soon extended to cover speaker identification, with safeguards similar to those used for police transcripts (Reference ChanChan, 2020).

As with forensic transcription, the law considers speaker recognition to be a topic of common knowledge. Again, however, while listeners confidently identify speakers in everyday situations, scientific studies show that this skill relies on context far more than listeners realise. When voices are heard out of context, as in a recording, recognition performance is surprisingly poor; more importantly, when voices are heard in a misleading context, they can be identified confidently but wrongly (see Reference Kreiman and SidtisKreiman & Sidtis, 2011: Chapter 7).

3.2.2 Early Use of Expert Witnesses

A further benefit of forensic audio was that it allowed investigators to commission expert assistance in profiling unknown speakers whose voices had been captured in audio. This approach had an early and dramatic start in Australia, with the infamous 1971 Qantas bomb hoax (the Wikipedia article ‘1971 Qantas bomb hoax’ provides a good introduction). A caller identifying himself as ‘Mr Brown’ left a message that a plane, in flight from Sydney to Hong Kong, had been planted with a bomb set to explode when the plane passed below a certain altitude. Authorities were unable to locate the bomb, or the bomber, so were forced to hand over a large cash ransom. After the ransom was paid, they were told there was no bomb: it had all been a hoax.

Police mounted a national manhunt. With no hard evidence, they explored every line of enquiry. Since some of the phone calls had been recorded, they asked a member of Sydney University’s English Department, Alex Jones, to assist in identifying the speaker’s accent. Jones reached the conclusion that the hoaxer was English, possibly from Cornwall.

Authorities enlisted the aid of Scotland Yard, but still the hoaxer could not be found – until his accomplice visited the local petrol station driving a series of expensive cars. Attendants reported him to police, and he quickly confessed his involvement, naming the mastermind as Peter Macari, an Englishman originally from Devon, now living in Australia. Macari initially denied involvement, but eventually police gathered compelling evidence against him, and he too confessed. This meant there was no trial, so Jones’ evidence was never tested in court – but the case brought his field to the attention of police, who called on him for other bomb threat cases, prevalent at the time, as well as for other types of analysis (see Reference JonesJones, 1994).

Sometimes Jones appeared for the defence. In one case, involving recordings of menacing telephone calls, he formed the view, on the basis of auditory analysis, that the suspect was not the speaker in the audio. To back up his opinion, he obtained a recording of the suspect reading the same words, and compared the voices with the aid of spectrograms (visual representations of speech, widely used in phonetics research). In the trial, the judge disallowed Jones’ evidence, as the courts did not recognise phonetics as a field of expertise. However, that decision was overturned when an appeal (R v Gilmore 1977) cited cases from the US where spectrograms had been admitted as ‘voiceprint’ evidence.

This raised protest from a police sound engineer, who published an article (Reference Hall and CollinsHall & Collins, 1980) pointing out that the voiceprint evidence used in US courts was a known pseudoscience. This is true (see IAFPA’s ‘Voiceprint Resolution’: iafpa.net/about/resolutions/). However, though Jones’ use of spectrograms would not be considered valid forensic phonetics today, it was not voiceprint evidence in the US sense. The terms ‘voiceprint’ and ‘spectrogram’ refer to the same thing, but the way they are used is very different (Reference HammarströmHammarström, 1987). Voiceprint methods were developed by engineers on the mistaken assumption that voices have unique characteristics, like fingerprints. Spectrographic analysis was, and remains, an important tool for responsible phonetic analysis of all kinds.

It is interesting to note that, despite their opposition to Jones’ evidence for the defence in Gilmore, police commissioned him to provide spectrographic evidence for subsequent trials, including that of the infamous ‘Woolworths bombers’ (R v McHardie & Danielson 1983), which is still cited as a precedent, though the methods are not approved by the current science of forensic phonetics.

3.2.3 A Growing Field

Though Alex Jones was the first linguist to provide speaker identification evidence in Australia, others gradually followed (see Reference EadesEades, 1994). Göran Hammarström, of Monash University, was an early practitioner (Reference HammarströmHammarström, 1987). In 1988, following Hammarström’s retirement, Heather Bowe established the Monash Speaker Identification Unit, through which Kate Storey and others undertook case work (Reference Bowe and StoreyBowe & Storey, 1995).

Around 1993, another infamous trial drew more phoneticians into forensics. Threatening phone calls, in fake Italian or Chinese accents, were received by NSW government and media offices, and office workers claimed to recognise the voice of local politician Barry Morris. Experts were instructed by both sides. While some provided evidence confirming the office workers’ views, Helen Fraser took a more cautious approach, for two reasons.

First, she noted that the office workers, as ear witnesses, were not reliable. To demonstrate this, she ran a small experiment in which listeners attempted to recognise voices of colleagues and family members without contextual information. To participants’ surprise, no one scored 100 per cent. Even for undisguised voices, most scored around 80–90 per cent, and confidence correlated poorly with accuracy. For disguised voices, scores were far lower, often at chance (in line with studies later reported in Reference Kreiman and SidtisKreiman & Sidtis, 2011). This made it likely that the office workers were not recognising Morris by voice alone, but were being influenced by contextual assumptions (see also Reference JonesJones, 1994).

Second, Fraser argued that phonetic science had not yet developed methods capable of reliably identifying speakers from their voices, especially with short samples of disguised speech. In the mid-1990s, the science of speaker recognition was in its infancy (see Reference NolanNolan, 1983; Reference Baldwin and FrenchBaldwin & French, 1990). Fraser sought to publicise the limitations of speaker recognition in police and legal newsletters – hoping that the information would enable the law to take a more realistic approach – and also presented her experience with the Barry Morris recordings in a special session at the 1995 IAFL conference at University of New England.

3.2.4 A Developing Science

Other phoneticians were experiencing similar concerns (cf. Reference Rose and DuncanRose & Duncan, 1995). By the mid-1990s, demand for forensic speaker identification expertise, previously sporadic, was burgeoning. This was due to an unintended side effect of the requirement that all police interviews with suspects should be recorded (Section 1.2.2). Where previously it had been rare to have an appropriate known sample of a voice to compare with a speaker heard in forensic audio, recorded interviews meant a known sample was now available for almost every case (cf. Reference FrenchFrench, 2017).

However, demand was not always being filled by properly qualified experts in phonetic science. In fact, despite the cases mentioned above, the courts often disallowed evidence from phoneticians in the belief that their expertise added nothing to the common knowledge already possessed by the jury. Some judges even preferred evidence from engineers, seeing this as more ‘scientific’ than phonetics.

Concerned by the poor quality of speaker recognition evidence, a group of ASSTA members established a Forensic Speech Science Standards Committee (FSSSC) aiming to develop standards to ensure that forensic speaker recognition was carried out by properly qualified experts. Founding members included John Ingram (chair), Phil Rose, Andy Butcher, Michael Wagner, Helen Fraser and Laura Tollfree.

To help upskill phoneticians in this new application of their science, Rose convened a 1996 seminar led by Francis Nolan of the University of Cambridge, one of the first phoneticians to study the phonetic bases of speaker recognition (Reference NolanNolan, 1983). ASSTA conferences started to have sessions on forensic phonetics, and several researchers and students started working on forensic speaker recognition (e.g. Reference Ingram, Prandolini and OngIngram, Prandolini & Ong, 1996). Also around this time, Laura Tollfree joined Monash University linguistics department, undertaking substantial, well-regarded research. However, while she inspired several students to take up forensic phonetics (see Reference LoakesLoakes, 2008; Reference Loakes and McDougallLoakes & McDougall, 2010), Tollfree herself left academia around 2000. Australian phonetician Duncan Markham, working overseas mainly on second language pronunciation, also contributed casework and research (Reference MarkhamMarkham, 1999).

3.2.5 The Paradigm Shift

A major development was sparked by Hugh Selby, then at the Australian National University Law School, who commissioned Phil Rose to write a chapter on forensic speaker identification (see Reference Freckelton and SelbyFreckelton & Selby, 2019). Selby gave Rose the first (1995) edition of Interpreting Evidence (Reference Robertson, Vignaux and BergerRobertson, Vignaux & Berger, 2016), which explained the need for the forensic sciences to use Bayesian statistics, especially Likelihood Ratio (LR). This important book had a powerful influence on Rose, who became a strong advocate for use of LR in speaker comparison (Reference RoseRose 2002), undertaking extensive research in this area and giving LR evidence in a number of high-profile cases (see https://philjohnrose.net).

Rose trained several graduate students in the use of LR, including Tony Reference AldermanAlderman (2005), Jennifer Elliot, Yuko Kinoshita and Shunichi Ishihara. He also brought Geoffrey Morrison from Canada as a research associate on a large ARC Discovery Project (2007–2010). Morrison then led an ARC Linkage project (2010–2013) through UNSW’s School of Electrical Engineering and Telecommunications, and later took a senior position at Aston Institute for Forensic Linguistics in England. From there he has been influential in gaining support from forensic scientists for use of LR in speaker comparison (e.g. Reference Morrison, Enzinger and HughesMorrison et al., 2021). Within forensic phonetics itself, however, though Rose’s Bayesian insights have been well accepted, it is fair to say that his highly technical and mathematical approach to using LR has been received with less enthusiasm (see responses to Reference Rose and MorrisonRose & Morrison, 2009; Reference Hudson, McDougall, Hughes, Knight and SetterHudson, McDougall & Hughes, 2021). This may be because, valuable as the concept of LR is, the fact that an expert report contains explicit calculation of an LR is not the only or even the main criterion distinguishing reliable from unreliable speaker comparison evidence. So even if all conclusions were given as a likelihood ratio, the courts could still be misled – for example if the LR is calculated on the basis of unjustified measurements or assumptions (Reference Fraser, Donohue, Ishihara and SteedFraser, 2012).

Another major problem, explored with some sensitivity by Reference RoseRose (2013), is the difficulty of explaining the mathematics of his LR approach, and the abstract acoustic analyses it favours, to the courts. As a result, these concepts have, to date, had little impact in reducing reliance on ad hoc experts for voice evidence in Australian courts – indeed this has expanded, with interpreters and translators now allowed to offer opinions regarding the identity of speakers of languages other than English (Reference ChanChan, 2020).

Further, outside the excellent, but rather technical, work of Rose’s group (e.g. Reference Kinoshita and IshiharaKinoshita & Ishihara, 2015), few Australian phoneticians have continued research, teaching or practice in forensic speaker comparison (though Kirsty McDougall has been a prolific researcher at the University of Cambridge, England). This means the pool of genuine experts available locally to give speaker comparison evidence is very small, with some indication that demand is being filled by less qualified personnel.

3.2.6 Looking Forward

Reference RoseRose’s (2002) book is rightly valued for bringing the crucial concepts of Bayesian reasoning to forensic speaker recognition. However, the focus on the mathematics of LR may have left the most important Bayesian insights obscure to some.

Perhaps these key insights can be clarified by considering why a Bayesian approach is so essential to forensic analysis in the first place. The forensic task is significantly different from ‘normal’ phonetic research, especially in the relationship between what is known and what is being determined. Normally, to investigate speaker characteristics, the analyst starts with known speakers, and studies how their voices vary depending on their known regional, social and other characteristics. In forensic casework, by contrast, the speaker’s identity is not known. It must be determined from speech characteristics observable in the audio.

The significance of this difference is highlighted by a useful analogy (cf. Reference KoehlerKoehler, 2013): if ‘normal’ phonetic analysis is like determining how likely it is that Jack is eating a meat pie, given that he is known to be at a football match, then forensic casework is like determining how likely it is that Jack is at a football match, given that he is known to be eating a meat pie. These two questions, though they may seem similar on the surface, are fundamentally different. Answering them requires not just different statistics but different methodologies. In particular, the second requires access to information not stated in the question, namely, how often Jack eats pies at venues other than football matches.

Turning the analogy to phonetics, consider, for example, the observation that a certain proportion of Melburnians pronounce ‘celery’ like ‘salary’ (see Reference Loakes, Clothier, Hajek and FletcherLoakes et al., 2014). Valuable as this information is, it cannot be used directly to determine the likelihood that a recorded speaker who pronounces ‘celery’ like ‘salary’ is a particular individual who is known to come from Melbourne. Among other necessary information, the analyst would need to know the proportion of non-Melburnians who pronounce ‘celery’ like ‘salary’; and the proportion of Melburnians who do not (consistently) use this pronunciation. These are complex questions. Answering them in the context of a particular case requires careful consideration of a range of factors spanning multiple branches of linguistic science. Importantly, for many of the factors, quantifiable information is not (yet) available.

Perhaps exploring this more conceptual understanding of Bayesian reasoning can give non-mathematicians better access to its insights, open up a broader set of research questions – and assist in developing accountable, evidence-based methods focusing on developing the capabilities of analysts as well as the discipline’s knowledge about speaker differences.

The overarching problem is that, as with forensic transcription (Section 3.1), no matter how good such evidence-based methods might become, as long as current legal procedures remain in place, good scientific methods alone cannot ensure that juries are provided with reliable assistance in determining the identity of speakers in forensic audio. Experts giving speaker comparison opinions are not always aware that a police opinion will likely have been provided to the court before the expert opinion is offered, and that the jury will be explicitly instructed to listen to the audio and reach their own conclusion as to which opinion they prefer – because speaker recognition is still considered by the law to be a matter of common knowledge.

It is essential to ensure that the law treats speaker recognition as a science, and acknowledges that well-established findings show many assumptions of common knowledge are simply incorrect. Equally, speech science must recognise that the law does not operate on scientific principles, and scientific evidence needs to be tailored to the realities of courtroom discourse (Section 2.1).

3.5.1 Expert Evidence/Casework

While linguists readily identify semantic and pragmatic meaning as linguistic phenomena, suitable for analysis with theories of language structure and use, judges and lawyers might be surprised to find that anyone other than a legally trained and qualified professional could shed light on the meaning of a legal text (Section 2.2.2). This provides the grounds for a contest of ontologies in relation to language and meaning. In law, there has been a reliance on fixed meanings and recourse to ‘dictionary definitions’ to resolve disputes over meaning, without taking account of context. This is described in Section 2.3.5 as the referential language ideology and underlies a number of misunderstandings about language by legal professionals.

In the current examination of meaning in written legal language, this ideology of referentialism appears to be expressed as a ‘turf war’, and the example provided below from case law as well as case studies from the authors’ own work demonstrate that referentialism is invoked differently by different courts – some judges holding that meaning is fixed regardless of context and therefore a linguist offers no special expertise, while some others find that the linguistic theory being applied in a linguistic report is too abstract and ignores context, thus rendering it irrelevant to the proceedings.

3.5.2 Other Applied Linguistic Work on Legal Language

As indicated in the introduction to Section 3.5, Australia is recognised internationally in the plain language and law movement, especially through the work of Australian linguist and Professor of English, the late Robert Eagleson, beginning in the 1970s. He was instrumental in the creation of the NRMA Plain English Insurance Policy in 1976. Eagleson’s ground-breaking report for the Law Reform Commission of Victoria (1987), Plain English and the Law, was republished online by the Victorian Law Reform Commission (2017) (see also Reference EaglesonEagleson, 1990). In 1990 Eagleson co-founded with Peter Butt the Law Foundation Centre for Plain Legal Language at the University of Sydney.

Reference GibbonsJohn Gibbons (1990, Reference Gibbons2001) worked with New South Wales police on simplifying the language of the police caution, and his report on this project shows that not all of his suggestions were adopted (Reference GibbonsGibbons, 1990, Reference Gibbons2001; see also Reference RockRock, 2007: 259 for a similar example from England and Wales). More recently, Gibbons has worked with the Victorian Charge Book Committee on making Jury Instructions more comprehensible. This work highlights the ongoing tension that arises from the two-audience problem of legal language (Reference GibbonsGibbons, 2003). Gibbons argues that it is important ‘not to sacrifice comprehension by the primary audience [here, the jury] for the sake of the secondary audience – an appeal court’ (Reference Gibbons2017: 144). With this approach, his analysis of linguistic and comprehension issues in jury instructions led to his suggestion for a ‘radical recasting’ of the wording of instructions rather than revising them ‘at a superficial linguistic level’ (2017: 159). But his suggestions for revision were not adopted by the committee.

Plain English is especially important in situations involving speakers of English as an additional language. In the last decade a number of linguists have worked with lawyers and legal organisations on improving access to justice through projects aimed at making legal language more comprehensible. These projects highlight ways in which the monolingual language ideology (Section 2.3.3) is being addressed, and show that linguists are making a significant and substantial contribution to such work.

The most accessible and extensive example is the Plain English Legal Dictionary: Northern Territory Criminal Law, jointly published by the independent Aboriginal research agency ARDS, the independent Aboriginal legal service NAAJA and the Northern Territory Government’s Aboriginal Interpreter Service AIS (Aboriginal Resource and Development Services et al., 2015). This dictionary takes a story-based approach to the collation of legal definitions in plain language into simple narratives showing common usage of these terms. More than 300 definitions are also provided in a regular, dictionary-style list. The main purpose of the dictionary is to improve access to justice for speakers of Australian Indigenous languages and speakers of Australian Aboriginal English. One of the challenges for the dictionary team is that many Anglo legal terms (e.g. ‘bail’, ‘arrest’ or ‘warrant’) have no direct equivalent in Indigenous languages, which are part of the oldest living culture in the world and do not include ways of expressing English settler legal concepts (see also Section 5).

A crucial part of the four-year preparation of the dictionary was the translation of legal terms into plain English before translation into Djambarrpuyngu (the main dialect of Yolngu Matha, a widely spoken Aboriginal language in the Northern Territory). Reiterative drafting was required between legal English, plain English and Djambarrpuyngu. The resulting Plain English Dictionary (ARDS et al., 2015) and Djambarrpuyngu Dictionary (ARDS, 2015) were the product of intensive collaboration between Aboriginal interpreters, and lawyers and linguists (including lawyer-linguist Ben Grimes and linguists Steve Swartz, Howard Amery and Marilyn McLellan).

The Northern Territory has also been leading the country in the preparation in languages other than English of recordings of suspects’ rights in police interviews (known as the caution). This began in the late 1990s with work by linguist Michael Cooke and lawyer Jenny Hardy to produce a plain English version, then approved by the Departments of the Attorney-General and the Office of the Director of Public Prosecutions, as well as police and representative lawyers. This ‘front translation’ (Reference 74BowenBowen, 2021b: 322) was then translated into fifteen Aboriginal languages. The resulting audio-recordings, intended as a resource for police, are described as ‘preambles’ to the caution (Reference MildrenMildren, 1999), providing explanation, rather than being limited to direct translation (Section 5.1; Reference 74BowenBowen, 2021b). There is no public information about the extent to which these recordings were used.

In 2015, the Northern Territory Aboriginal Interpreter Service completed a similar project, in order to provide much shorter versions, again working from a plain English front translation. Audiovisual recordings were produced in eighteen Aboriginal languages of suspects’ right to silence and their right to inform a person of their location: https://cmc.nt.gov.au/aboriginal-affairs/aboriginal-interpreter-service/aboriginal-language-police-cautions-aboriginal-interpreter-service. Lawyer-linguist Ben Grimes was part of the multi-disciplinary group comprising interpreters, linguists, police officers and lawyers. Again, it has not been possible to ascertain the extent to which these recordings are used. See Section 5.1 for Reference 74BowenBowen’s (2021b) discussion of the complex issues involved in these ‘intercultural translation[s] of vague legal language’ (p. 308).

A similar project in Western Australia was part of the police response to the decision in the Gibson case, to be discussed in Section 3.6.1. The ‘Yarning’ app, on the phones of all police officers, delivers information, in plain English and eight Aboriginal languages, about the right of suspects in police custody to contact a lawyer or family member. The app helps police identify a suspect’s strongest language, and it also provides a quick link to Aboriginal Interpreting Western Australia (AIWA) who worked with police to develop the app (Reference BowenBowen, 2021c). Lawyer-linguist Alex Bowen was part of AIWA’s team for this project.

A different approach to plain English in the law, with which lawyer-linguist Ben Grimes has been involved, is the Blurred Borders Resource kit. This is a legal communication tool used by frontline service providers working with Aboriginal clients in regional and remote locations of Western Australia and the Northern Territory. Focusing on the key legal concepts around bail and the criminal process, and family violence and child protection, this kit uses visual art, plain language and storytelling to explain legal concepts and processes, https://blurredborders.legalaid.wa.gov.au/about-blurred-borders.

Leading Australian lexicographer Pam Peters has developed a free online dictionary of 350 words and phrases (known as a ‘term bank’) used in Australia Family Law and mediation services. Commissioned by the Federal Attorney-General’s Department, the TermFinder™ term bank in family law has also been translated into eight community languages, namely Arabic, Chinese (Modern), Chinese (Traditional), Filipino, Korean, Spanish, Turkish and Vietnamese: https://lawtermfinder.mq.edu.au/.

A pilot Indigenous family law term bank in Easy English has also been created, which provides visual support, and is organised by related concepts, rather than alphabetically, partly using resources from the Blurred Borders project (see above). https://lawtermfinder.mq.edu.au/easy-read-home.php. There are plans to extend the current 150-word term bank, and to provide audio-recordings of all the verbal material. As with the Plain English Criminal Law Dictionary, and the Blurred Borders project, an initiative to help Indigenous people understand the law has the potential to also explain the law to a wide range of other linguistically and culturally diverse people, as well as people with cognitive disabilities.

Attention to legal language is also one of the characteristics of the Law Society of the Northern Territory’s (2015) Indigenous Protocols for Lawyers, an initiative by lawyers and for lawyers, which seeks to codify knowledge from applied linguistics and language work into recommendations and guidelines. One of the six Protocols is ‘Use “plain English” to the greatest extent possible’. The publication’s five-page explanation and exemplification of how to do this has been adopted in several other publications for lawyers and courts, including JCCD (2022) and Law Society of South Australia (2020). Lawyer-linguist Ben Grimes was actively involved in the writing of this document.

Finally, the Guidelines for Communicating Rights to Non-Native Speakers of English in Australia, England and Wales, and the USA, www.aaal.org/guidelines-for-communication-rights, produced by the Communication of Rights Group (CoRG, 2015), comprised seven Australian forensic linguists as part of the group of 21 linguists, lawyers, psychologists and interpreters: Michael Cooke, Diana Eades (co-convenor), John Gibbons, Ben Grimes, Sandra Hale, David Moore and Ikuko Nakane. The focus of CoRG was the communication of rights to non-native speakers in police interviews with suspects, and the intention was to ‘translate’ research on this topic for practitioners and policy makers. Drawing on linguistic and psychological research, as well as the collective experience of working with specific cases, this Communication of Rights Group produced a 2000-word guidelines document setting out seven recommendations, as well as providing an explanation intended to be accessible for police officers, lawyers, judges and justice administrators. The document’s Appendix is a bibliography of relevant research (Reference Eades and PavlenkoEades & Pavlenko, 2016).

While these positive developments in the use of plain, or clear, English cannot be disputed, the CoRG (2015) Guidelines point out that ‘there is no one formulation of rights/cautions that would be immediately understandable to all’. In his work on translations of the right to silence in the Northern Territory, Reference BowenBowen (2019, Reference Bowen2021a, Reference 74Bowen2021b) goes further, showing how linguistic work on the wording of legal language is not enough when there are gaps in contextual knowledge about the culture of the law (see Section 5.1).

3.5.3 Research

Australian research on the language of the legal process has paid less attention to the structural aspects of the language of law than the pragmatic aspects (Section 4). But some linguistic research has addressed discourse structure and semantics.

Yon Reference MaleyMaley (1994) used a systemic functional linguistics approach to overview the three legal discourses of legislation, trial proceedings and judicial judgments. Her conclusion, in the early years of linguistic research in the law, that reform of legal language is limited by institutional structures, practices and ideologies, resonates with contemporary linguistic concerns about the limits of intercultural translation (e.g. Bowen, Section 5.1).

Three Australian linguists have researched the use of a particular semantic theory (Natural Semantic Metalanguage) in exploring and explaining meaning in law: Cliff Goddard with the words ‘enterprise’, ‘reckless’ and ‘sudden’ (Reference Goddard1996); lawyer-linguist Ian Langford (2000) with a focus on the legal terms ‘murder’, ‘manslaughter’ and ‘homicide’; and Anna Reference Rañosa-Madrunio and MartinWierzbicka (2003) on the legal terms ‘reasonable man’ and ‘beyond reasonable doubt’.

Reference HeydonHeydon (2005) analysed the use of legal terminology in police interviews by applying Conversation Analysis and interactional sociolinguistics to recordings of police interviews from Victoria. She found that the word ‘charge’, that has a common meaning of ‘payment owing’, is easily confused for the legal meaning of ‘charge’, meaning that a person is identified by police as accused of committing an offence, particularly, as in the case described, when the charge is for ‘damage to the door’. This use of ‘legal homonyms’ (Reference TiersmaTiersma, 2000) occurs regularly in police interview data, and the impact on laypeople noted by Reference HeydonHeydon (2005) is similar to the concerns raised in relation to legal language for partial speakers of English in research (Aboriginal Resource and Development Services, 2008).

3.6.1 Second Language Speakers: Expert Evidence

After Strehlow’s 1959 evidence in the Stuart case (Section 1.3.1), the earliest expert linguistic evidence involving languages other than English was presented by John Gibbons in several cases in the 1980s and 1990s, before the introduction of audio-recording of police interviews with suspects (Section 1.2.2). Gibbons used findings from research on second language acquisition and language testing in answering questions about the accuracy of alleged confessions which non-native speakers had claimed were fabricated (Reference GibbonsGibbons, 1990, Reference Gibbons1995; see also Reference JensenJensen, 1995; Reference EadesEades, 1993 for a second dialect speaker).

Over a period of more than two decades, several linguists have written expert reports, and in some cases given expert evidence, concerning linguistic resources of individual non-native speakers of English, typically addressing the question of whether a defendant could have understood their right to silence, and/or particular questions, in a police interview. The monolingual language ideology (Section 2.3.3) is at the heart of these cases (as indicated by both discussions with the linguists named below and the publications referred to). Linguists often report the need to address the lack of awareness in the law that even if speakers of English as an additional language can conduct a basic conversation in English, they may nevertheless need an interpreter for more complex interactions such as a police interview (see also Section 3.6.3).

There is no way of knowing how many linguists have provided such expertise, but we are aware that a list would include the following, some of whom have published in the linguistics or law field, drawing on a specific case in which they have presented expert evidence: Denise Angelo, Alex Bowen, Michael Cooke (e.g. Reference Cooke1996), Maria Doyle, Diana Eades (e.g. 2018), Cathie Elder, Rod Gardner, John Gibbons (e.g. 1990, 1995), Helen Harper, David Ingram, John Ingram, Marie-Thérèse Reference JensenJensen (1995), Ute Knoch, Ian Malcolm, Prudy McLaughlin, David Moore, Ikuko Nakane and Carmel O’Shannessy.

Several of these cases, and much of the research discussed in this section, have involved Aboriginal second language speakers of English. In contrast, very little work has dealt with the other group of Indigenous Australians, namely Torres Strait Islanders. However, in an unreported 1995 case in north Queensland, Helen Harper gave expert linguistic evidence about the police interview of a Torres Strait Creole speaker charged with attempted murder (R v Izumi 1995).

Part of Harper’s evidence was that the accused had not properly understood the police caution or a number of the questions she was asked by police (see also Section 3.5.1). Two of the main issues in Harper’s linguistic evidence were (i) that Torres Strait Creole is not the same as English, and (ii) that there was no evidence that Ms Izumi had understood the caution. The judge accepted Harper’s evidence and decided to exclude the police interview evidence. The murder charge was dropped, and Mrs Izumi pleaded guilty to unlawful wounding (Reference TresizeTresize, 1996).

In WA v Gibson 2014, a Western Australian Supreme Court judge made important rulings about language proficiency and the need for an interpreter, in a refreshing departure from the referentialist and monolingual ideologies so frequently seen in the law (Section 2.3). Mr Gibson is a speaker of three Aboriginal languages and a partial speaker of English (Section 3.6.3). The judge refused to accept that his ‘yes’ answers in a police interview to ‘do you understand?’ questions were indications of understanding (referring to the expert linguistic evidence provided by Eades, as well as David Ingram and Maria Doyle; see Reference EadesEades, 2018). Instead, the judge pointed to problems with the ‘assumption that the answers given always reflect an understanding of the questions asked’ (paragraph 74). Not only did the judge reject the idea that the meaning of the answer ‘yes’ was fixed, he also made it clear that linguistic and cultural contexts were central to meaning and to his decision in the case: particularly Mr Gibson’s very low English proficiency and the pressure in his first language (from an older male relative to whom he owed respect) that directly contradicted what the police were saying to him in his third language.

3.6.2 Languages Other than English: Research on Interpreting

In Australia more than 300 languages other than English are spoken, and 21 per cent of Australians speak a language other than English at home. The conservative and traditionally Anglocentric institution of the law has come a long way in the last fifty years in recognising the importance of interpreting, but linguistic research, casework and interactions with legal professionals and law enforcement show that the monolingual language ideology (Section 2.3.3) continues to influence policy and practice.

For example, the law’s assumption that monolingualism is, and should be, the norm can give rise to objections about a second language speaker’s request for an interpreter, as indicated above. Such objections can be seen in terms of a common misconception among legal professionals that if such a person can converse in English at a conversational level, then they do not need an interpreter, and even that they should not have an interpreter in legal interviews (Section 3.6.3; Reference EadesEades, 2018). However, research on the language of bilingual people in a range of countries and social contexts shows that it is common for people to have much less proficiency in their second language than their first language.

Frequent misconceptions about the nature of bilingual language use can sometimes result in situations where people with some bilingual fluency are expected to act as if they are monolingual speakers of their language other than English. In courts, witnesses who do not hide the fact that they have some bilingual fluency can run the risk of being seen as misrepresenting their need for an interpreter. Thus, in order to deal with complex questions with an interpreter, they are expected to ignore the simple English questions (for which they may not need interpreting assistance), as the court expects them to act as monolingual speakers of their first language. In legal situations which do not adhere to this aspect of the monolingual language ideology, the rights to justice of bilingual people may sometimes be better served, as Reference CookeCooke (1996) illustrates with an example of a Djambarrpunyngu and English-speaking defendant from northern Australia (R v M 1995).

The most prolific Australian linguistic researcher on issues impacting users of languages other than English in the law is Sandra Hale, whose contributions are recognised not just within Australia, but internationally, since her 2004 study of court interpreting in trials involving Spanish interpreters. Hale’s micro-analysis of lawyers’ discourse markers and question types in interpreted proceedings shows how and why it is so difficult for interpreters to achieve one of the central goals of interpreting, namely pragmatic accuracy (that is, conveying speaker meaning, such as sarcasm, surprise and disbelief, as well as utterance meaning).

Since this seminal study, Hale and her colleagues have continued to research a wide range of aspects of legal interpreting, including issues of quality, codes of ethics, role expectations, the impact of interpreters on legal proceedings, the perceptions of interpreters held by service providers and non-English speakers, and the impact of working conditions on interpreter performance (e.g. Reference Hale and SternHale & Stern, 2011; Reference HaleHale, 2013). The findings of this research directly impact the training and certification of interpreters, areas of professional practice in which Hale is arguably the national leader. At the same time Hale has played a key role in the education of legal professionals, including judicial officers, about what is involved in interpreting and how they can work with interpreters in the delivery of justice.

Using a combination of experimental and discourse analytical methods, Hale’s most recent research has focused on different aspects of police and court interpreting, including the effect on interpreter performance of interpreter education and background (Reference Hale, Goodman-Delahunty and MartschukHale, Goodman-Delahunty & Martschuk, 2019), language combination, interpreter mode (Reference Hale, San Roque, Spencer and NapierHale et al., 2017) and interpreter presence or remoteness – e.g. via videolink– (Reference Hale, Goodman-Delahunty, Martschuk and LimHale et al., 2022). Interpreters’ maintenance of strategic investigative interview techniques, including rapport building, as well as their interpretation of vulgar and emotional language has also been a key area of investigation (Reference Hale, Martschuk, Goodman-Delahunty, Taibi and XuHale et al., 2020; see also Reference Lai and MulayimLai & Mulayim, 2014; Reference Mulayim, Lai and NormaMulayim, Lai & Norma, 2015).

In all of her work Hale counters widespread misconceptions within the law of what it means to speak English as an additional language, especially the language ideologies of referentialism, monolingualism and communication as coded message transfer (Section 2.3). Consistent with these assumptions about language is the widespread misconception among legal professionals that interpreters can, and should, act as a language ‘machine’ or conduit (Section 2.3.3.; see also Reference Smith-KhanSmith-Khan, 2017).

Hale’s direct input was crucial to the publication of the Recommended National Standards for Working with Interpreters in Courts and Tribunals (JCCD, 2022; and see Section 2.3.3). Throughout this document, accurate interpreting is defined, following the AUSIT Code of Ethics (AUSIT, 2012), as

This important document makes it clear that the job of interpreting is to transfer both semantic and pragmatic meaning, that is referential meaning and speaker’s meaning in the context of their talk.

While most of Hale’s research on the actual process of legal interpreting has focused on pragmatic challenges in court contexts, a number of other linguists have examined other kinds of linguistic challenges for legal interpreting, or for lay participants in legal processes who speak English as an additional language but do not have full proficiency in English.

Ludmila Stern (e.g. Reference Stern2001) examines linguistic and cross-cultural challenges in the three war crimes prosecutions and the trial held in Australia in the late 1980s to early 1990s, when several Ukrainian people were brought to Australia as witnesses to crimes allegedly committed during the second world war. She compares these challenges and related interpreting issues with trials in the International Criminal Tribunal for the Former Yugoslavia in The Hague. Jieun Lee (e.g. Reference Lee2009) highlights linguistic differences between the Korean and English languages in the expression of definiteness in noun phrases, presenting a compelling example which highlights the importance of interpreters being seen as language professionals who may sometimes need to explain linguistic challenges to the court.

While most research on interpreters and second language speakers in the legal system has focused on courtroom hearings, Nakane has examined interpreted police interviews, using a discourse pragmatic framework (e.g. Reference Nakane2007, Reference Nakane2014). The catalyst for this research was her analysis of a police interview in a 2002 Federal Court case, and her expert report for an Individual Communication submitted to the United Nations Human Rights Commission in 2005, concerning a 1994 Victorian County Court case. Nakane’s work highlights the impact of interpreters on the interactional dynamics of police interviews, as interviewer and interviewee negotiate over the construction of competing stories.

Reference MacFarlane, Kurt, Heydon and RohMacfarlane et al. (2019) take a critical lens to the institutional management of interpreters in police interviews and courts in Australia. The authors note that a failure to address shortfalls in the provision of a professional interpreter service reflects the institutional racism in Australia’s legal system whereby English proficiency is privileged and overrides the preferences of clients who speak other languages, including Aboriginal languages. Despite Reference Lo BiancoLo Bianco’s (1987) optimism about a shift away from entrenched English monolingualism, this work finds that ‘while the country might claim to tolerate limited amounts of linguistic diversity in private circles, efforts to enforce linguistic unification through English remain structurally embedded, especially in public domains’ (p. 52).

A somewhat different issue concerning interpreters in the legal process is the wide-ranging Australian linguistic research concerning interpreters for deaf jurors. This research is leading the way internationally in investigating whether the exclusion of deaf people from juries, common in many countries, is justified by the reasons advanced within the law.

More than fifteen years ago three Macquarie University researchers began investigating this issue (linguist Jemina Napier, lawyer David Spencer and interpreter-linguist David Sabolcec), producing a report for the New South Wales Law Reform Commission (Reference Napier, Spencer and SabolcecNapier, Spencer & Sabolcec, 2007). Since then, linguists Napier and Hale, and lawyers Spencer and Mehera San Roque, have conducted research including studies which reveal that there is no difference in the ability of hearing jurors and deaf jurors (with interpreters) to understand judges’ directions and summations to the jury. One of the other studies in the project surveyed legal professionals and sign language interpreters in seven countries about their perceptions of the impact of a deaf person (receiving the evidence and participating in jury deliberations with the assistance of an interpreter) on the jury and on the outcome of trials. This study was further enriched by focus group research with stakeholders. The final study (e.g. Reference Hale, San Roque, Spencer and NapierHale et al., 2017; Reference Napier, Spencer and HaleNapier et al., 2019) was based on a mock trial involving eight legal professionals, two deaf jurors and two professionally qualified sign language interpreters.

The conclusion of this multi-faceted research programme is that ‘deaf people can comprehend legal discourse found in the judicial system and that their presence in the court and jury rooms is not detrimental to the delivery of a fair trial to the accused’ (Reference Napier, Spencer and HaleNapier et al. 2019: 263). In addition, the research highlighted the ‘substantive and logistical hurdles’ that need to be overcome for deaf people to serve on juries, and suggests education programmes and ‘systematic changes in the legal system’ (Reference Napier, Spencer and HaleNapier et al. 2019: 263, see also Reference Hale, San Roque, Spencer and NapierHale et al., 2017).

A hallmark of this research is the involvement of lawyers and linguists in the research teams and the inclusion in its many stages of legal professionals, interpreters and deaf people. Some parallels can be noted with the way in which linguists have worked with legal professionals and Aboriginal interpreters to address problems arising from legal language for Aboriginal second language speakers of English in the Northern Territory and Western Australia (Section 3.6.3). And as with this latter work, the research on deaf jurors is creating change within the legal system. Australian Human Rights organisation Remedy Australia (2021) shows ‘Australia is on the cusp of making its juries more inclusive for people with hearing and vision disabilities’ (p 19), with the Australian Capital Territory leading the way with legislative reform.

3.6.3 Aboriginal and Torres Strait Islander Second Language Speakers specifically

The specific linguistic contexts for speakers of Aboriginal languages in the legal process have been written about by linguists since the late 1970s (e.g. Reference BrennanBrennan 1979; Reference NashNash, 1979; Reference McKayMcKay, 1985). Michael Walsh is one of the linguists who has published on linguistic (and interactional) issues involved in taking evidence from Aboriginal people in land claims and native title proceedings (at both tribunals and trials). Harold Reference Koch and PrideKoch (1985) analysed linguistic features (‘pronunciation, vocabulary and grammar’) of the non-standard English of Aboriginal claimants which lead to miscommunication during the preparation of land claims and tribunal hearings.

Michael Cooke’s research in remote Arnhem Land since the 1990s (e.g. Reference Cooke1995a, Reference Cooke and Morris1995b, Reference Cooke1996) provides the most detailed examination of miscommunication with Aboriginal speakers of traditional languages in the legal process. His work, as well as David Moore’s (e.g. Reference Moore, Wakabayashi and O’Hagan2022), with speakers of Aboriginal languages, adds to that by Hale and others discussed in Section 3.6.2 on interpreting for speakers of immigrant languages.

However, in contrast to most of the Australian research with non-native-speaking immigrants, most of the research and expert evidence with Aboriginal people who are non-native speakers of English focuses on legal processes where interpreting has not been provided, because participants (whether lay participants and/or legal professionals) assume that English will work. There are difficulties in theory and in practice with deciding when an interpreter is required to communicate adequately with an Aboriginal person in a legal context (Law Society of the Northern Territory, 2015; Reference CookeCooke, 2002).

Cooke’s work (see references above, and Reference Cooke2002, Reference Cooke2004, Reference 75Cooke2009) has highlighted the linguistic reality that many Aboriginal people from remote communities are partial speakers of English as a second (or subsequent) language, with their capacity to communicate in standard English compromised because of interference from their Aboriginal language(s), at all levels from pronunciation to pragmatics. Frequent and often unrecognised miscommunication commonly results, particularly at lower levels of English proficiency (Reference CookeCooke, 2002: 8–11).

Cooke’s contributions include his 1996 case study (see Section 3.6.2) which highlights the importance of stand-by interpreting in court for witnesses (including defendants) who have sufficient proficiency in English to answer simple biographical and narrative questions, but whose proficiency is not sufficient to talk about more complex issues such as explaining their state of mind. Cooke has also provided expert evidence in cases in NT and Western Australia, and has made a major contribution to the application of linguistics in the training and accrediting of interpreters over several decades.

Reference CookeCooke (2002), Nick Reference Evans, Henderson and NashEvans (2002) and Reference BowenBowen (2019) have also shown some specific linguistic challenges for Aboriginal partial speakers of English in police interviews, courtroom hearings and land claim hearings which relate to the mismatch between specific English modal verbs (‘have to’, ‘should’ and ‘can’) and the expression of modality in Aboriginal languages. While this might appear to be a small area of linguistic difference in the resources of people communicating in English in legal contexts, the work of these scholars reveals the important consequences for justice. It also highlights the widespread impact of the lack of understanding of the need for interpreters that results from the monolingual language ideology (Section 2.3.3).

Linguists working with second language speakers of Aboriginal languages in the law have not restricted their attention to the linguistic context that is the focus of this section. See Sections 4 and 5 for studies and applications that focus on interactional and sociocultural contexts.

3.6.4 Second Dialect Speakers: Aboriginal People

Some attention has also been focused on linguistic issues impacting people who speak Aboriginal English as their main language, for example in relation to miscommunication that can arise from dialectal differences in word choice and meaning (e.g. Reference 76EadesEades, 2010: 173–5). For research and expert evidence about differences in language use and sociocultural context for Aboriginal speakers of varieties of English, see Sections 4.2.2, 4.2.3 and 5.

3.6.5 Developmental Issues: Child Witnesses

Finally in this consideration of linguistic resources, research by Reference Brennan and BrennanBrennan and Brennan (1988; see also Reference BrennanBrennan 1995) addresses the consequences of ignoring the developmental linguistic proficiency of young children when they are questioned in court. Their study first categorised thirteen linguistic features of complex or confusing questions asked of child witnesses. They then tested thirty school children on the comprehensibility of questions asked by lawyers, compared to those asked by teachers and counsellors. The study found that the children were unable to adequately understand 85 per cent of the lawyer questions, while almost all of the counsellor questions and most of the teacher questions were understood most of the time.

Reference HeydonHeydon (1998, Reference Heydon2005: 148–64, Reference Heydon, Kredens and Gozdz-Roszkowski2008a) and Reference La Rooy, Heydon, Korkman, Myklebust, Oxburgh, Myklebust, Grant and MilneLa Rooy et al. (2015) have examined the language of questions used in interviews with children by police. This work has added important nuance to the literature on child interviewing, such as further specification of appropriate question types beyond the usual ‘open vs closed’ typology.

Pamela Snow has contributed research on narratives used by young speakers that supports speech pathologists in working with children and young people in detention, and in other institutional justice settings (e.g. Reference Snow and PowellSnow & Powell, 2005).

In the thirty-five years since Brennan’s study, the law has taken many steps to address a range of issues (not limited to linguistic issues) impacting legal interviews with children. This work includes pre-recording the direct evidence of child witnesses for the trial in a less intimidating environment than the courtroom and reducing the time the child needs to spend answering questions in court. A recent initiative is the provision of a ‘Children’s Champion’ (or ‘Witness Intermediary’) to assess communication needs of child witnesses and to inform police and the court of the best ways to communicate with the child giving evidence.

4.1.1 Interactional Structures in Investigative Interviews

Linguistic contributions to the work of police investigators and other forms of investigative process in the Australian justice environment have focused on the interactional structures of interviews and how such structures affect the information that is provided by the lay participants in legal settings. Police investigative interviews form the basis for information gathering in many criminal justice cases. Such interviews are conducted according to regulations, which are explained in police training manuals, written to ensure that police interviewing procedures comply with the relevant state and federal legislation.

However, beyond the specific requirements of the legislation, which focuses on legal rights and points of proof, police are trained in strategies of questioning and conversational management techniques. Linguistic research has identified the impact of various question types on the informativeness of the response, and the extent to which turn-taking strategies can affect the cooperation of the interviewee – an interactional version of rapport building.

The capacity for interviewees in various legal contexts to provide their own version of events is explored through Goffman’s participation roles, as well as conversation analysis and topic management (Reference HeydonHeydon, 2019). The linguistic analysis of the interactional context in the justice system provides insights that complement other forms of analysis, mainly psychology (Reference HeydonHeydon, 2019). For instance, psychologists have identified the ways that an eyewitness’s brain stores the memory of an event, and the cognitive factors that can affect recall (Reference Milne and BullMilne & Bull, 1999), whereas linguistic analysis can identify the most appropriate conversational strategies to maximise the opportunities for that recall to be expressed verbally in response to a question or prompt (Reference HeydonHeydon, 2012).

The nature of the work by Australian linguists on police interviewing has been significant in its consideration of current police training methods and strategies employed by officers when conducting interviews to maximise the evidentiary value of the information provided by the interviewee. Thus, the institutional setting has been important to the examination of the interactional contexts of police interviewing.

The interactional and institutional contexts can also affect the gathering of information in a written format. Research into the interactional features of written forms to collect reports of sexual assault as an alternative or precursor to a formal police interview is being led by an Australian research team with a linguistic focus.

Again, the intertwining of institutional and interactional contexts for forensic linguistic research has led to innovative work on the nature of narratives submitted as part of an online account of sexual violence and the impact on policing and justice outcomes for stakeholders (Reference Heydon and PowellHeydon & Powell, 2018).

The multilingual context of legal interactions in Australia has provided a wealth of opportunities for the exploration of interpreter-mediated communications. Cross-cultural communication as well as pragmatic analysis of interpreted police interviews has contributed substantially to the field internationally. For instance, Miranda Lai’s work on the relationship between cognitive interviewing strategies and interpreted interviewing was the first of its kind (Reference Heydon and LaiHeydon & Lai, 2013; Reference Lai and MulayimLai & Mulayim, 2014; Reference Mulayim, Lai and NormaMulayim, Lai & Norma, 2015) and has gained traction internationally, especially amongst practitioners. For example, the FBI and the High Value Detainee Interrogation Group were specifically interested in the work of Australian linguists (including Heydon) in 2013 because of the unique findings that Australian researchers had generated in relation to police interviewing in multilingual contexts.

4.1.2 Lie Detection

As discussed above, the focus of police interviewing in Australia has followed professional practice in the UK and New Zealand, which emphasises the elicitation of a free narrative as the most reliable form of information gathering. There is a further benefit to this narrative-based approach to police interviewing, and other forms of legal information gathering, which is its role in deception detection.

Credibility assessment is a common aspect of proceedings in courtrooms in Australia and around the world. The right to a fair trial rests in part on the principle that a person who is accused of, or witness to, a crime must be given the opportunity to give their evidence in person where their ‘credibility’ can be assessed by the judge and/or jury. This is a complex area of jurisprudence, but underlying the notion of credibility is a belief that live, spoken interaction can provide an opportunity for a hearer to assess the truthfulness, or credibility, of the speaker, based on the hearer’s perception of the speaker’s demeanour, absent any relevant factual evidence (Reference Porter and Ten BrinkePorter & Ten Brinke, 2009: 120). This reliance on hearer perception is challenged by research, which demonstrates that indicators of truthfulness commonly used in such situations (eye contact, fidgeting, hesitant speech) are culturally bound, or equally associated with nervousness, and do not stand up to scientific testing (e.g. Reference Granhag, Vrij and VerschuereGranhag, Vrij & Verschuere, 2015). Additionally, Australian research has pointed out that factors which have been found to sway a jury’s opinion include the performance of a court interpreter, which can be highly prejudicial for witnesses and defendants providing evidence in a language other than English (Reference Hale, San Roque, Spencer and NapierHale et al., 2017).

In police investigations, lie detection has largely given way to investigative processes that rely on hard evidence and detailed accounts from suspects or witnesses. However, among the most high-profile consultant investigators engaged by Australian police there are still proponents of scientifically invalid techniques such as statement analysis (searching written statements for ‘markers’ of deception), polygraph testing and ‘micro-expressions’, a behavioural analysis technique which sits within the ‘body language’ category of lie detection (see for instance https://www.stevevanaperen.com/).

As a form of language analysis, the lie detection methods that rely on the examination of written statements for markers of detection have been critiqued in Australia for their lack of an underlying theory of language structure and use. Reference Heydon, Smets and VrijHeydon (2008b, Reference Heydon2019) finds that purported lie detection techniques such as Scientific Content Analysis (SCAN) are not based on sound linguistic theory and are promoted to police as almost magical in their power to uncover deception in a written statement. In a system of interrogation which relies on coercion to elicit confessions, police are more likely to be drawn to anything that can assure them that they are interrogating only the guilty parties. Hence, in the USA where the coercive Reid technique of interviewing is dominant, lie detection courses and techniques find a ready market. In Australian police academies, lie detection has largely been replaced by training that emphasises an investigative approach to interviewing (Reference Adam and van GoldeAdam & van Golde, 2020). However, lie detection proponents still find an audience in Australia amongst other investigative and interview-based professions, such as in the recruitment and security sectors.

Australian police interviewing contexts are dominated by the PEACE model of interviewing, which relies on cognitive psychology for its underpinning conceptual framework and evidence base. The techniques used within this model promote the elicitation of a free narrative, and this provides a better opportunity for lie detection than the aforementioned methods, like SCAN, polygraph testing or micro expression analysis. The reason for this is that a detailed oral account from a suspect, or witness, freely given without leading or suggestive questions is regarded as the most reliable form of evidence. It also maximises the opportunities for detectives to check the interviewee’s account against any other evidence gathered in the case. The linguistic analysis of participant roles provides an interactional perspective on the elicitation of information that can separate information that originates from the interviewee’s own version of events from information that originates from a police version of events. This could be valuable in the assessment of interviewee credibility when testing the interviewee’s version against known facts in the case.

4.2.1 Early Work

The first Australian linguistic publication about language in the courtroom was Reference EadesEades’ (1992) handbook for lawyers, written to dispel widespread myths and ignorance about Aboriginal ways of speaking English, and to suggest ways lawyers could engage in more effective communication with Aboriginal clients and witnesses.

This publication was partly motivated by judicial reactions to Eades’ expert linguistic evidence in the appeal in R v Condren 1987 (Section 1.3.2). The evidence was about the striking differences in an alleged confession to police between the words attributed to an Aboriginal man charged of murder, and his dialect of Aboriginal English. The linguistic evidence was not accepted by the court, whose reasons included biological determinism as an explanation for language usage (in effect, that the Aboriginal man’s mother was not dark-skinned enough, R v Condren 1987: 275), as well as linguistically uninformed views about language variation.

The earliest linguistic research on courtroom interactions (Reference WalshWalsh, 1994) arose from Michael Walsh’s fieldwork in several Aboriginal communities, mostly in remote areas of northern Australia. Walsh found that conversation in these traditionally oriented communities was mostly non-dyadic and continuous, with people often spending time together with the communication channel ‘switched on’, but including periods of silence, and where talk is not necessarily directed to a particular interlocutor. He showed the contrast with mainstream Australian interactional style, and particularly in the courtroom, where talk is almost always dyadic and non-continuous, and where talk is ‘strictly regimented’ (p. 226).

4.2.2 Questions and Answers

The central role of the question-answer structure in the courtroom (and throughout the legal process) has been a focus of all the linguistic writing on courtroom language. Reference LuchjenbroersLuchjenbroers (1997) contributed to the growing international findings about how defendants and witnesses are controlled, coerced and manipulated through the rigid and asymmetrical discourse structure of courtroom hearings, which restricts the interactional rights of witnesses to providing answers to specific questions (see also Reference GibbonsGibbons, 2003). Reference EadesEades (2000) questions the assumption in much of this work that the syntactic form of question types in court has a fixed relationship with the way in which power is exercised, with a study in a NSW court showing how supposedly controlling question types can be taken by witnesses in some hearings as an invitation to explain. And Reference Maley, Sarangi and CoulthardMaley’s (2000) analysis of expert witness testimony in a case in the Land and Environment Court makes the case that linguistic analysis needs to include more than ‘surface or formal properties’ of the talk, such as sentence form or turn-taking.

Reference EadesEades (e.g. 2000, Reference Eades, Fraser, Siegel, McNamara and Baker2003, Reference Eades2008b), Reference GrantWalsh (2008) and Reference CookeCooke (e.g. 1996, Reference Cooke2002) have reported on problems, for Aboriginal witnesses in particular, that arise from the reliance on the question-answer structure.

In Reference EadesEades’ (2000) study, lawyers and judges appear to be overly preoccupied with individual answers to questions rather than the story which a witness is trying to present to the court. The study shows how a focus on answers to questions can prevent lawyers and judges from realising how much they fail to understand about some of the Aboriginal witnesses and their evidence. This lack of understanding resulted, in this New South Wales study, not primarily from linguistic difference, but from differences in sociocultural context (Section 5), compounded by limiting linguistic practices, here courtroom discourse structure.

4.2.3 Assumptions about Storytelling and Retelling

At the centre of courtroom procedure are culturally based assumptions about how witnesses’ accounts/stories should be structured, retold and evaluated. Reference EadesEades’ (2008a, Reference Eades2008b, Reference Eades2012) in-depth analysis of the cross-examination of three teenage Aboriginal boys in Brisbane Magistrate’s Court (Crawford v Venardos 1995) is the first Australian linguistic work to come to terms with these assumptions, or language ideologies (see Section 2.3). Her work draws on that of linguistic anthropologists and others in the field of language ideologies in the law (e.g. Reference MatoesianMatoesian, 2001; Reference TrinchTrinch, 2003; Reference EhrlichEhrlich, 2012). Eades’ study highlights the following assumptions in law concerning verbal interaction, when witness’s accounts are told and retold (by themselves and others):

Eades’ micro-analysis of the cross-examination in this case shows the operation of these language ideologies in the courtroom ‘business as usual’ to legitimise the actions of police removing Aboriginal children from the street (without any evidence of the children’s wrongdoing and without charging them with any offence) and abandoning them in the middle of the night in an isolated industrial wasteland. And this language in court is also shown to play a role in reinforcing public discourse about Aboriginal children as a threat to public safety.

These four language ideologies at the heart of the way that laypeople’s stories are heard, shaped and assessed in court can be seen as part of the referential ideology introduced in Section 2, that is, the metacommunicative norm that ‘we can examine individual utterances out of context’ (Reference Ford and MertzFord & Mertz, 2016: 4). They also reveal the assumption that the ‘structure of language in court is neutral and reveals the truth’ (Reference Ford and MertzFord & Mertz, 2016: 5, drawing on Reference MatoesianMatoesian, 2016). The referential language ideology is taken up further by Reference Ehrlich and EadesEhrlich and Eades (2016) in their consideration of linguistic and discursive evidence in a range of legal settings, which shows how the law ignores linguistic, situational, interactional and sociocultural context in its determination of questions about consent.

4.2.4 Assessing the Stories of Asylum Seekers

These language ideologies are also an important part of lawyer-linguist Laura Smith-Khan’s research on communication with and about asylum seekers, which is resulting in extensive publications in both legal and linguistic journals (e.g. Reference Smith-Khan2017, Reference Smith-Khan2019a, Reference Smith-Khan2019b, Reference Smith-Khan2022). At the heart of the refugee status determination (RSD) process is the asylum seeker’s story –– of home country, persecution or fear of persecution, often war, and escape. When the veracity of an applicant’s story is assessed, by government officials and judicial officers in the tribunal and the court which decide appeals against an initial negative decision, apparent inconsistencies in the applicant’s various tellings and retellings of the story become crucial.

Smith-Khan focuses primarily on the analysis of institutional texts, especially on decisions of courts and tribunals which consider asylum interviews (and accompanying application forms and other evidence), as well as published government policy documents which guide the interviews and the RSD process. Her examination (e.g. Reference Smith-Khan2019b) of these institutional texts exposes the monolingual language ideology (Section 2.3.3), for example in expectations that a bilingual person should be fully proficient in both (or all) of the languages they speak. And the four language ideologies from Eades’ work outlined above also impact the way that decision-makers assess the witnesses’ account. (Unlike approaches used in some criminal investigations which rely on false beliefs about ‘markers of deception’ (Section 4.1.2), official guidance to the tribunal decision-makers cautions them – but only in general terms – to be aware of ‘the effect of cultural differences on demeanour and oral communication’ (Reference Smith-KhanSmith-Khan, 2019b: 413.)

Building on earlier findings of co-construction of accounts/stories, Smith-Khan shows not only how asylum seekers’ stories are constructed, conceptualised, evaluated and contested, but also that credibility itself is also co-constructed, through interviews and policies, in which these problematic assumptions about language play a role. Her most recent work gives reason for optimism that the use of sociolinguistic approaches which limit reliance on problematic language ideologies can lead to the situation where credibility assessments would come to ‘play a less prominent role in RSD’ and asylum seekers ‘would be more likely to be given the benefit of the doubt’ (Reference Smith-Khan2022: 14).

4.2.5 Other Legal Settings

Australian researchers have also contributed to the small body of linguistic research in alternative legal settings. Reference Candlin, Maley, Gunnarson, Linell and NordbergCandlin and Maley’s (1997) analysis of family law mediation sessions found that they combined features of both therapeutic discourse and legal discourse. Zappavigna, Dwyer and Martin (2016; see also Reference Zappavigna and MartinZappavigna & Martin, 2018) conducted ethnographic research on 18 Youth Justice Conferences in Sydney. These conferences are a restorative justice alternative to court for children (under 18) who admit to an offence.

Using functional linguistic and semiotic analysis, they uncover the discursive processes by which interaction in the conferences, strongly directed by the (trained) conference convenors, works to achieve the transformations at the heart of the goals of this form of restorative justice. Their study details convenors ‘scaffolding’ the negotiation of participants’ feelings, and often encouraging participants (including the young offender, the victim and family and support people of each) to ‘reframe the matter they are dealing with as an infringement of family, religious, or cultural values, and not simply as a legal violation’ (Reference Zappavigna and MartinZappavigna & Martin, 2018: 290). This in-depth linguistic study contributes to the wider sociolegal field by providing the first analysis in the world of the discursive activation of reintegrative shaming, which has been at the heart of the theory of restorative justice.

Reference StroudStroud (2010) describes processes in the Koori Court in Victoria, an example of an Indigenous sentencing court, where Elders and other respected people from local Indigenous communities sit with the magistrate to talk with defendants who have pleaded guilty, before the magistrate decides on the sentence. As these Indigenous courts are an addition to the sentencing process, they are not bound by rules of evidence, and witnesses speak directly, without lawyers. Defendants’ personal, family and social situations are examined in a ‘power-sharing’ legal context in which local cultural beliefs, lifestyles and practices (including ways of communicating) are recognised as central to the issues which need to be addressed in relation to the offender’s actions, relationships and sentencing (see Reference 76EadesEades, 2010: 225–8).

Reference Dieckmann and Rojas-LizanaDieckmann and Rojas-Lizana (2016) use discourse analysis to study free legal advice meetings, in particular looking for evidence of possible power imbalance between lawyers and clients. Focusing on the use of legal or formal register and interruptions, they found that the lawyer advice meetings studied are characterised by the ‘discourse of facilitation (participatory) rather than of domination (authoritarian) on the part of the lawyer’. Thus, they align more with expert advice-giving sessions than with studies (in other countries) of lawyer-client consultation.

Finally, in addition to her work on Australian police interviews, Ikuko Nakane has been researching the ‘hybrid’ Japanese courtroom (e.g. Reference NakaneNakane, 2020). Her main focus is the ways in which the combined adversarial and inquisitorial approaches in trials are realised in courtroom discourse.

6.3.1 In Legal Proceedings

Beginning with the Queensland criminal case of R v Aubrey 1995, the work of Diana Eades on communicating with Aboriginal people in the law has been cited in several cases where no expert linguistic evidence has been provided (see Appendix). In another such case, R v BL 2015, in addition to citing Eades’ work, the judicial decision also cites the work of Michael Cooke, and the Guidelines for Communicating Rights to Non-Native Speakers of English in Australia, England and Wales, and the USA (CoRG, 2015, Section 3.5.2).

The most important case to cite linguistic research is Nguyen v R 2020 in which the High Court of Australia (our highest court) decided an issue involving fairness in the prosecution of people ‘with cultural and linguistic disadvantage’ (with no specific reference to Aboriginal people). Justice Edelman’s judgment includes the adoption of Eades’ work about the difficulties involved in giving an account of events (or telling and retelling your story, Section 4.2.3) within the constraints of police interview and courtroom questioning, referring to Reference EadesEades (2012). This judgment marks a new development in the awareness by lawyers and judges of linguistic research, particularly because there was no expert linguistic evidence in this case. It also marks a step forward in our legal system’s understanding and recognition of the consequences of linguistic and cultural diversity.

6.3.2 In Commissioned Reports, Judicial Benchbooks and Similar

Michael Cooke, Sandra Hale and Georgina Heydon have all produced reports commissioned by legal organisations, as indicated in the Appendix. The Appendix also provides evidence of reports and other publications produced for judicial officers by researchers and support staff of courts and tribunals that have drawn on the work of Michael Cooke, Diana Eades and Sandra Hale, for more than two decades.