Paper Presentation Abstracts

(Alphabetical Order by Surname)

Sune Aken (University of Copenhagen ~ Denmark)

Genre Research and Forensic Linguistics: An Invitation to a Dialogue
The paper offers an invitation to consider academic collaborations between contemporary genre research, as it has developed over the last 30 odd years, and forensic linguistics. To open up this line of inquiry the paper presents some of the fundamental assumptions of genre research, including some of the work being done on juridical genres, and connect them tentatively with topics in forensic linguistics.

Gaby Axer (University of Bonn ~ Germany)                                     
A Cross-linguistic Study of Authorship Features in Instant Messaging

Authorship analysis is one of the major fields in Forensic Linguistics and has been implemented as evidence in various cases, often in connection with text messages. However, most work has focused on native speakers writing in their L1, or, in cases of L2 involvement, on native language identification (NLID). This paper sets out to examine whether a closed set of authors show a relation between their discriminant features of authorship in their L1 (German) and L2 (English) writing. For this purpose, three native speakers of German provided one instant messaging conversation conducted in each language. These are anonymised and split into known and questioned texts for a blind study: First, an authorship analysis within the English corpus is carried out during which the texts of undisputed authorship are analysed for individual features in order to select potentially distinctive markers of authorship between the three authors. These are then applied to the questioned texts to come to a conclusion with regards to their authorship. Second, the same procedure is repeated in the German corpus separately. Third, the discriminant features are compared between the two language sets with the goal of exploring the possibility of pairing the two sets which belong to the same author. This paper reports on the most promising findings regarding the cross-linguistic features of authorship found in the corpus and discusses the main obstacles in data gathering and coding, as well as their potential influences on those results.
                                                 
Stilistische Analyse von Autorenmerkmalen in WhatsApp Nachrichten
Die Autorenanalyse stellt eines der großen Gebiete der forensischen Linguistik dar und kommt in unterschiedlichen Formen in Ermittlungen zum Einsatz, in zunehmendem Maße auch in elektronischen Medien, wie beispielsweise SMS und den diese ablösenden mobilen Sofortnachrichtendiensten (instant messenger) wie WhatsApp und Threema. Für eine Blindstudie von sprachübergreifenden Autorenmerkmalen in WhatsApp-Nachrichten wurde folgendes Korpus erstellt: Drei Muttersprachlerinnen des Deutschen (L1) mit hoher Englisch-Kompetenz (L2) stellten je eine authentische WhatsApp-Unterhaltung auf Deutsch und eine auf Englisch zur Verfügung. Diese wurden zunächst anonymisiert und für die Analyse in bekannte (known/K-texts) und fragliche (questioned/Q-texts) Subkorpora unterteilt. Mithilfe einer stilistischen Analyse werden potentiell idiosynkratische Merkmale herausgearbeitet, um auf deren Grundlage die in diesem geschlossenen Set diskriminierenden Merkmale zu erarbeiten (Phase 1). Dies geschieht auf Basis von Konsistenz-Kodierung und schließt neben der stilistischen Analyse auch die Fehleranalyse mit ein. Die fraglichen Subkorpora werden nun auf diese diskriminierenden Merkmalsgruppen hin untersucht und jedes Merkmal in Hinblick auf Konsistenz mit jeder der drei bekannten Autorinnen codiert. Auf Grundlage dieser Codierung erfolgt eine Zuweisung der fraglichen Texte zu den bekannten Autorinnen, gefolgt von einer Evaluation der Ergebnisse. Dieses Paper konzentriert sich auf die Analyse der deutschen Daten und geht näher auf die Herausforderungen der Analyse und methodische Aspekte ein.

Susan Blackwell (Vrije Universiteit Amsterdam ~ the Netherlands)           
‘Let her be sworn’: child witnesses in the Old Bailey Corpus

Common-law legal systems have traditionally regarded children as unreliable witnesses, prone to fantasy and suggestibility. Accordingly, rules of evidence which apply to child witnesses often include features such as a competency test, corroboration of evidence and a caution from the judge to the jury. These features have only recently been discarded in the legal system of England and Wales. This paper reports a systematic study of the judicial treatment of child witnesses whose evidence is recorded in the Old Bailey Corpus, a searchable digital archive of the The Proceedings of the Old Bailey from 1674-1913 (www.oldbaileyonline.org). The record shows numerous instances of quite young children giving evidence as defendants, victims and witnesses in serious cases, covering offences ranging from highway robbery to rape, where the possible penalties included death or deportation. However, young witnesses were not always permitted to testify. The judges seem to have gone to some lengths to establish the age of any child who appeared before them and to satisfy themselves that the witness understood “the nature of an oath”, which was typically confirmed in the religious language of the time. Although the reports cannot be taken as verbatim, the Old Bailey records suggest that those children who were allowed to testify were rightly regarded by the Court as credible witnesses, often able to stand their ground under rigorous or even aggressive cross-examination about traumatic events.

Tanya Karoli Christensen (University of Copenhagen ~ Denmark)
Developing a Format for Police Interview Summaries: A Status Report

Early studies of the effects of audio taping police interviews showed benefits to interview time, information gathering, confession rates, etc., while—curiously—only few officers, prosecutors and defense attorneys reported listening to the recordings because of time concerns (Willis et al. 1988; Baldwin & Bedward 1991). Conversely, written reports are condensed and digitally searchable, but suffer from severe risks of introducing officers’ bias onto the statements (Fisher 1993). In this talk, I present the interim results of a pilot project detailing the requirements and challenges related to summarizing recorded police interviews. A joint effort of teachers and students at the Danish Police Academy and the University of Copenhagen (UCPH), the project consists of three phases: 1. Recordings of police interview training sessions; 2. Summaries of selected recordings by forensic linguistics students; and 3. Evaluations of summaries by police cadets, detectives and prosecutors. For the recently completed phase 2, FL students were instructed to summarize one of two interviews according to either 1) a reconstructed chronology of events, or 2) the witness’ own narrative structure, which is often cyclical in consisting of several re-tellings with different foci, as guided by police questioning (Rock 2001; Labov & Waletsky 1967).Current concerns of the participants include the information loss inherent in transforming spoken data to written form (Haberland & Mortensen 2016), developing standardized ways of reporting witnesses’ epistemic stance (e.g. Kärkkäinen 2003) and employing transcription linking software (Hazel et al. 2012) to facilitate direct (time-stamped) access from summaries to corresponding sections of the recordings.

Grace Sullivan Buker and Belen Lowrey (Aston University ~ England)
Talking Back, Talking Black: An Investigation of Escalation within the Arrest of Sand

On July 10th, 2015 a young African American woman, Sandra Bland, was stopped by Officer Brian Encinia for failing to signal a lane change. Although it began as a routine traffic stop, it quickly escalated into shouting, physical confrontation, and Bland’s subsequent arrest. This arrest sparked public outrage as it occurred amidst a number of police-civilian interactions in which the police apparently escalated into aggressive behavior against unarmed African American civilians. This paper examines the transcript of the Bland arrest and analyzes it with respect to the Use of Force Continuum, the guidelines as to how much force law enforcement officers may use against a resisting subject (NIJ 2009). According to the literature (Stetser 2001; Desmedt 2009; NIJ 2009), this continuum is a model used to clarify for law enforcement officers and civilians alike the rationale behind the use of force. Our paper finds the ways in which the law enforcement officers utilize imperatives (Portner, 2012), certain modal structures (Rubinstein 2012; Portner 2016), and explicit mention of legal consequences (Phillips 1982; Agar 1985; Heritage 1997) does not follow the standards of the continuum. Furthermore, motivated by an analysis from Holliday, Burdin, and Tyler (2015) of Bland’s arrest, we expand upon a discussion on how Bland’s incredulous responses may have been considered “noncompliant” (e.g. talking back) from the outset, thus allowing for escalation, and was a result of linguistic ideologies and racial profiling.

Patricia Canning (Utrecht University ~ The Netherlands)
Police Collusion in the construction of witness narratives after the Hillsborough Football Stadium Disaster

On 15th April 1989, at the FA semi-final match between Nottingham Forest and Liverpool FC at Hillsborough, a series of institutional failures led to a fatal crush in two spectator pens behind the goal. Ninety-six people died in what has become known as the ‘Hillsborough Disaster’. Within minutes, the South Yorkshire Police (SYP), government officials, and the media constructed a ‘story’ linking fans’ so-called ‘hooliganism’ to the cause of the disaster (Hillsborough Independent Panel Report, 2012). Despite no supporting evidence, ‘hooliganism’ became a salient feature within SYP witness statements, manifested through frequent – and often incongruent – references to fans’ alleged alcohol consumption and disorderly behaviour (HIP archive). My paper argues for police collusion in the evidence-gathering process of the Hillsborough investigation in its analysis of linguistic micro-features that include lexical and syntactic patterns, as well as ‘macrosemantics’, or ‘topics’ (van Dijk, 1993), in a small sample (n=12) of key witness statements. In collaboration with survivors of Hillsborough, this study proposes that the SYP established ‘sock-puppet’ witnesses comprised of local residents who collaborated with police to support the institutionally constructed macro-narrative of fans’ hooliganism, ultimately blaming fans for the disaster. This paper is very timely; the Independent Police Complaints Commission (IPCC) recently completed its investigation into allegations of criminal behaviour by the SYP (and others) and submitted files on 8 (unidentified) individuals to the Director of Public Prosecutions (DPP). It is hoped that this research will contribute to the ongoing fight to bring those who suppressed the truth to justice.

Biyu (Jade) Du (Newcastle University ~ England)
When English is used as a lingua franca: communication challenges in court interpreting

This paper attempts to investigate the communicative complexity between the accused and interpreters when they use English as a lingua franca in Chinese courtroom. The study is set against the backdrop of globalisation and migration in the past two decades when China has experienced rapid economic growth and radical social changes, attracting a large number of overseas visitors. When non-Mandarin speakers come into contact with the Chinese legal system, they rely on interpreters to participate in the legal proceedings. English is the most commonly used language by foreign litigants with criminal charges in Chinese courtroom. Drawing upon a four-month fieldwork in three courts of a Chinese city, and on the basis of data analysis of the recordings of seven criminal hearings that mostly involve defendants who speak English as the second language, this paper reveals the communication challenges facing interpreters and defendants owing to their different ‘varieties of English’ – differences in pronunciation become obstacles to mutual intelligibility. Adding to the difficulty in achieving mutual understanding is disparate legal culture and legal systems. These thus have an impact on interpreters’ performance in terms of accuracy and completeness, potentially placing non-Mandarin-speaking litigants to a situation where they can neither fully understand courtroom interaction nor make their defence understood by others. This study is to shed light on an understanding of the complexity in court interpreting and to raise awareness of importance of training interpreters to work with second-language-speakers of English.

Annie Heini (Aston University ~ England)
A Comparative Analysis of Police Interview Discourse in Interviews with 17- and 18-Year-Old Suspects

English law sees a line between children and adults, i.e. between 17- and 18-year-olds. This division means different police guidelines for suspect interviews with both age groups. Persons up to 17 are entitled to special measures; when they turn 18, these measures cease. This overnight change in treatment can have detrimental effects on 18-year-olds; however, special measures intended for children can also be deemed inappropriate for 17-year-old interviewees. This paper presents preliminary findings of my PhD research, which investigates the discourse of interviews with 17- and 18-year-old suspects. Interviews are preponderantly linguistic events and it is by examining discursive features that we uncover exactly how the age-line affects the interaction. The research question of how age-specific police guidance and legislation influence language is answered by analysing data from either side of this legal divide using qualitative approaches based on Conversation Analysis. Specifically, conversational features including turn-taking, back-channelling, pause lengths and overlaps are examined. My research makes clear the arbitrary nature of the line between childhood and adulthood, and shows how this affects interview discourse. Linguistic research into this age-line has never been conducted, and the findings identify the most salient issues, resulting in implications for practice and building a solid foundation for future research.

Daniel Leisser (University of Vienna ~ Austria)
The Participation Gap in the Normative Space: The State of Play in Austria

It is doubtful whether the normative space is a participatory space. The language of the law is a language of institutional power, which is evident in the divergence between ordinary language interpretation and institutionalised meaning-making processes. Assuming that linguistic indeterminacy plays a crucial role in the identification and analysis of juridical power, this chapter provides a multifaceted discussion from a legal linguistics perspective, elaborating on theoretical as well as empirical aspects (tenant law and criminal law) relating to the relationship between legal language use and the concept of participation. It is argued that the participation gap in the normative space may be understood as co-determined by the lack of educational measures in Austrian secondary schools. The integration of awareness building towards legal language use could be achieved by establishing basic legal competence in the relevant national curricula. By tackling the current lack of legal competence in Austrian classrooms, one may achieve a reduction of the participation gap. Unless sufficient steps are taken in education to enable students to approach normative systems with more scrutiny, individuals will not be treated as legal subjects, but will instead continue to be subjected to the rule of law.

Ulrike Lohner (University of Hamburg ~ Germany)
Evidentielle Adverbien im Deutschen: eine empirische Untersuchung in der Krimi-Serie Tatort

Anders als z.B. im Quechua ist Evidentialität im Deutschen nicht (stark) als grammatische Kategorie grammatikalisiert (s. aber Diewald/Smirnova 2010). Dennoch kann evidentielle Funktion mit bestimmten, u.a. lexikalischen Mitteln ausgedrückt werden. Hierunter zählen Modaladverbien wie offensichtlich, anscheinend und angeblich. Während zum Englischen mit Glougie (2016) bereits eine umfangreiche Untersuchung evidentieller Adverbien vorliegt, fanden sie im Deutschen bislang wenig Beachtung. Der Vortrag präsentiert die Ergebnisse einer empirischen Untersuchung von evidentiellen Adverbien in polizeilichen Vernehmungen. Hierzu wurden interrogative Gesprächssequenzen zwischen Ermittlern und Zeugen/Verdächtigen aus Dialogtexten der deutschen Tatort-Serie auf die Nutzung solcher Adverbien hin analysiert. Im Fokus der Untersuchung standen vierzehn Adverbien, deren evidentielle Verwendung in Bezug auf den Evidenztypen bereits etymologisch suggeriert wird (z.B. offensichtlich: etwas ist offen sichtbar). Datengeleitet wurden die einzelnen Adverbien Evidenztypen zugeordnet, die aus der bestehenden Forschungsliteratur bekannt sind (z.B. Matthewson 2015). Es wird deutlich, dass im Deutschen eine systematische Zuordnung von Adverbien zu Evidenztypen besteht. Die Auswertung der Daten zeigt unter anderem, dass angeblich und scheinbar reportative, anscheinend, offenbar und wahrscheinlich inferentielle Evidentialität zum Ausdruck bringen. Offensichtlich wird demnach für Informationen verwendet, deren Herkunft entweder direkt visuell oder aber die Schlussfolgerung aus unmittelbar erlebten Ereignissen ist. Das so gewonnene bessere Verständnis von Funktion und Bedeutung evidentieller Adverbien im Diskurs kann für die Schulung polizeilichen Personals genutzt werden und damit sowohl Qualität und Effizienz als auch die inhaltliche Interpretation von Vernehmungen und Befragungen entscheidend verbessern.

Karoline Marko (University of Graz ~ Austria)
“On the Advice of my Attorney, I Refuse to Answer that Question”: An Analysis of Body Language in the Kristin Smart Case

On May 25th, 1996, 19-year-old Kristin Smart disappeared from the campus of her university in San Luis Obispo. Despite delays in the investigation a main suspect, Paul Flores, was soon discovered. He was the last to be seen with Kristin the night she disappeared. Although some evidence indicates that Paul Flores killed Kristin Smart, the evidence was never enough to initiate proceedings against him. Based mainly on Navarro (2016), but also Dimitrius and Partick Mazzarella (2008), and Hartley and Karnich (2007), available material from Flores’ police interview in 1996 and his deposition in 1997 are analyzed for body language. Flores exhibits two different kinds of non-verbal behavior in his two interviews: while he talks and gestures much in the first one, he sits still and does not answer questions at all in the second one. Further, his non-verbal behavior will be contrasted with Melvin de la Motte’s, his lawyer, who is present in the 1997 deposition. Is it possible that non-verbal behavior and body language disclose details about the case that the suspect chose to remain silent about?

Iman Nick (Germanic Society for Forensic Linguistics ~ Germany)
A Forensic Onomastic Investigation of Criminal Aliases Used by 100 Fugitives Featured on the FBI Most Wanted List

According to recent statistics from the US FBI, just over 5.6 million criminal offenses were registered in 2013. Of these crimes, ca. 80.35%, were committed by so-called “known offenders”. Despite this fact, establishing a reliable, positive match between offenders and crimes is complicated by the fact that offenders often assume multiple names. This mixed method investigates the criminal aliases used by 100 fugitives on the FBI Most Wanted List. Using statistical analyses punctuated by authentic offender profiles, this forensic onomastic study will demonstrate the aliases which criminals select follow identifiable onomastic patterns both within-speaker and crime-type.

Criminal Aliases in the Lone Star State: An Empirical Forensic Onomastic Case-Study of 150 Texas Arrestees
According to the Texas Department of Public Safety, in 2014, ca.600,000 men age 17 and older were incarcerated. Their crimes ranged from vandalism and drug abuse to sexual assault and first-degree murder. Despite this variation, there was one feature which the apprehended shared: a criminal alias. This forensic onomastic case-study of 150 captured Texas male-offenders presents several strong, clearly discernible, generative patterns found in the criminals’ monikers. Then, information is given on same time, within-group onomastic variations found across the offender crime-type and official demographic classification. The potential implications and applications of this research for law enforcement will be discussed.

 

E pluribus unum: Hate Mail after the Presidential Election of Donald J. Trump
On the 20th of January 2017, Republican candidate Donald J. Trump was officially sworn in as President of the United States of America. Despite the fact that the Democratic candidate Hillary Rodham Clinton had won the popular vote, the majority of the electoral college cast their vote for Trump. This political division was a fateful harbinger for what some observers have called an unprecedented wave of post-election hate crimes. In this investigation, hate mail messages were analyzed using a psycholinguistic template developed on the basis of the FBI-BAU [Behavioral Analysis Unit] protocol. The results of these analyses revealed clear patterns in the Anonymous Threatening Communications.

Ria Perkins (Aston University ~ England)
Politeness strategies and Native Language Influence Detection: the benefit of using explanations in NLID

This paper demonstrates the benefits of developing interlingual explanations for native language influence detection (NLID). It presents findings from a sub-project analysing politeness and the role it can play in Native Language Influence detection. Many existing studies take a predominantly computational approach to NLID, and do not seek to understand the linguistic reasons why certain features occur. This paper discusses the benefits of using interlingual (and intercultural) explanations, and demonstrates that such an approach serves to make the analysis much stronger. It focuses specifically on politeness strategies, and how these are done differently in the relevant languages and cultures, and demonstrates how this can support and deepen NLID analysis in casework situations. The data comprises four corpora of blogs and webfora by native speakers of; Russian, Polish, Chinese, and Korean. Most of the existing literature on interlanguage takes a pedagogical perspective and therefore predominantly investigates student data. By contrast, the research reported here is grounded within forensic linguistics and focuses on case relevant data, specifically several corpora of weblogs and webfora. This paper pays particular attention to the practical implications of the findings reported, and examines the methods and applications for casework. It demonstrates how an understanding of politeness strategies can aid NLID, and considers how the wider findings can be applied to potential real life cases involving NLID or forensic authorship analysis.

Isabel Picornell (QED Limited ~ England)
Linguistically Profiling the Trump Russia Dossier

In January 2017, BuzzFeed News published online a collection of memos totalling 35 pages, written in English over 7 months in 2016, alleging that the Russian government had been cultivating close ties with the then US Presidential candidate (now US President), Donald Trump. Since the online publication, controversy has surrounded the authorship of the 17 memos. The source of the memos was subsequently identified as a UK native-English speaker working in the Intelligence industry. However, to date, the individual has not confirmed or denied personally writing all or any part of the memos, and various newspaper articles have questioned his role in their authorship. The Trump Russia Dossier presents the ultimate challenge for forensic linguistics as everything about it is challenged: its authorship, its context, and its authenticity. Using different approaches to authorship analysis, this study throws new light on the linguistic origins and context of the Trump Dossier by attempting to answer standard authorship questions such as: are the texts a product of single or multiple authorship; what inferences can be made regarding the linguistic background of the author(s); and is the language in the texts consistent with the claimed context.

Linda Pfister (University of Uppsala ~ Sweden)
Die Rechtfertigung des Wohles von Flüchtlingskindern in schwedischen Präzedenzfällen

Der schwedische Staat hat sich dazu verpflichtet, Kinderrechte in Asylprozessen uneingeschränkt zu gewährleisten. Die Präambel des Ausländergesetzes schreibt vor, das Kindeswohl werde immer berücksichtigt, auch wenn es nicht allein ausschlaggebend für die Entscheidung über einen Asylantrag sei. Dass jedoch der Migrationsobergerichtshof meist zum Nachteil für Flüchtlingskinder urteilt, könnte gegen dieses grundlegende Prinzip verstoßen. Ziel meines Doktorandenprojektes ist es zu untersuchen, wie das Kindeswohl in den Präzedenzfällen dieser letzten schwedischen Migrationsinstanz während der letzten zehn Jahre sprachlich gerechtfertigt wurde. Zu diesem Zweck gründet sich die Studie auf der systemisch-funktionalen Grammatik [1] – einer Theorie über die Fähigkeit von Sprache, in verschieden Diskursarten Bedeutung zu realisieren. Der soziosemiotische Erklärungsrahmen ermöglicht die Annahme, Entscheidungsträger nähmen eine aktive Rolle in geltendem Recht ein, die in den Beschlüssen des Präzedenzrechts sichtbar wird [2]. Der Forschungsschwerpunkt liegt auf den asymmetrischen Machtverhältnissen, die zwischen asylsuchenden Kindern und verantwortlichen staatlichen Behörden (wieder)erschaffen werden. Welche Bedeutungen und Interessen werden bei der Rechtfertigung des Kindeswohls konstruiert und verbalisiert? Kommt der Präambel des Ausländergesetzes gleiches Gewicht zu, unabhängig vom Ausgang des Urteils, dem Jahr der Entscheidung und der Verfügbarkeit von Erziehungsberechtigten? Diese und ähnliche Schlüsselfragen sollen die Rolle der Entscheidungsträger als aktive Handlungsträger im Präzedenzrecht hervorheben und somit auch die Wichtigkeit linguistischer Forschung im angewandten Recht unterstreichen.

Sune Sønderberg Mortensen (Roskilde University ~ Denmark)
Questioning Style in Danish Criminal Trials

A great deal of research in courtroom interaction has focused on question types and questioning regimes characterising the common law adversarial system of trial (e.g. Danet et al. 1976; Atkinson & Drew 1979; Harris 1984; Bülow-Moller 1992; Cotterill 2003; Gibbons 2008; Harris 2011). Studies of e.g. Danish and Swedish courtroom interaction observe how the casual and non-confrontational atmosphere of Scandinavian courts significantly deviates from e.g. that of US courts (Adelswärd et al. 1987; Jacobsen 2002; Kæraa 2011; Mortensen & Mortensen 2017), despite employing a similar adversarial participant structure (e.g. Anderson 1992). Yet, these differences are still to be explored linguistically. In this paper, I will examine the questioning style, in terms of syntactic structure and pragmatic and interactional functions, in three Danish misdemeanour-level trials, based on transcribed audio recordings from a Copenhagen court. Central features of the questioning style will be discussed in relation to US courtroom questioning data.

James Tompkinson (University of York ~ England)
“They’ve been watching too many action films”: Lay-listener Voice Descriptions in Evaluations of Threatening Communications

The majority of earwitnesses to crimes have no prior linguistic training (Griffiths, 2012), yet there are many cases where those who commit crimes are heard but not seen (Nolan and Grabe, 1996:74). One such example is when anonymous bomb threats are made via the telephone. In the UK, the government issues a bomb threat checklist, designed to elicit information about an unknown threatener’s voice. This paper examines how linguistically untrained listeners describe a speaker’s voice using such a checklist, and assesses the linguistic robustness of this practice. Across three experiments, participants were instructed to describe the voices of different speakers producing utterances in both familiar and unfamiliar languages. The familiar language stimuli were indirect bomb threats, while in the unfamiliar language, half of the participants were primed to believe the recordings were bomb threat calls made to emergency service operators. Results showed that some aspects of voice, including pitch and certain voice qualities, were identified with a relative degree of accuracy. However, listeners also showed willingness to infer information about speakers’ emotional state, body size, sincerity to commit harm and perceived level of threat from vocal characteristics alone. The results support the view that there may be some usefulness in gaining lay-listener descriptions of voices in anonymous bomb threat cases, but that linguists should assist by providing modifications to current practice to ensure that linguistically informative descriptions are obtained. Recommendations for the improvement of the current UK bomb threat checklist are also considered in this paper.

Fleur van der Houwen and Marja Kramp (Vrije Universiteit Amsterdam ~ the Netherlands)
Telling and Eliciting Personal Experiences and Motives from LGBT Asylum Seekers in the Netherlands

Using a conversation analytic approach, this study examines the different ways in which LGBT asylum seekers tell IND (Dutch Immigration and Naturalization Service) officials about personal experiences and motives for seeking asylum as well as how these are elicited. This exploratory study is part of a larger research project which examines interpreter mediated asylum procedures. Asylum seekers are normally interviewed twice and it is in the second interview that the asylum seekers tell about their motives for leaving their country of origin. The corpus consists of the audio and written transcripts of 23 interviews at the Dutch Immigration with people who seek asylum for being LGBT, 10 of which are second interviews.

Edina Vinnai (University of Miskolc ~ Hungary)
The Fulfilment of Right to Information in Hungarian Legal Practice

The right to information is a crucial element of the right to fair trial, which is highlighted in several EU legal acts. In case of criminal proceedings, Directive 2012/13/EU requires both information on procedural rights and also Letters of Rights designed for detained defendants to be provided in ‘simple and accessible language’. This intention proves that the right to fair trial has a legal as well as a linguistic aspect. But what makes an oral or written information on certain rights and obligations in a legal procedure ‘simple and accessible’? In the first ‘law and language’ research project in Hungary, carried out between 2000 and 2003 by researchers of the University of Miskolc, delivering information on rights and obligations by police officers and judges was, among others, examined. During the research, we prepared voice recordings at police interrogations and court hearings and then transcribed them. From 2014 until 2018 another research is being carried out and new voice recordings are prepared. Based on the database stemming from these projects, the presentation reviews the former and the present Hungarian practice on giving information by professional participants to lay persons. I will examine the following questions: in what language styles and in what manner interrogators and judges inform the lay persons about their rights and obligations; how it is recorded in the official record; what the non-professional participants of the legal procedures understand from the information they were given, etc.? As regards written information provided for detained defendants, I will shortly present the main conclusions of the Country Report issued by the Hungarian Helsinki Committee. In 2015, in the framework of the project titled ‘Accessible Letters of Rights in Europe’ it was examined whether the official Hungarian Letter of Rights was “simple and accessible” for non-lawyers.

Dakota Wing (Hofstra University ~ United States)
The Progression of Interpersonal Stances in the Writings of School Shooters

A recent school shooting in France and the “223 school shootings in America since 2013” (Everytown For Gun Safety, 2017) speak to the prevailing global nature of school shooters. Existing research surrounding this global phenomenon primarily come from psychological and sociological perspectives that rely on hindsight, hearsay, and self-report (Langman 2009b, O’Toole, 2000; Vossekuil, 2002, White, 2017). Using Appraisal Analysis (Martin and White, 2005), this study examines the authentic language data of six American high school shooters at seven time periods within the year preceding their shootings, revealing patterns of interpersonal stances over time. It was hypothesized that there would be an increase in positive attitudinal markers towards the self, an increase in negative attitudinal markers towards others, and an increase in negative attitudinal markers of the shooters’ opinion of others’ opinion of himself. Additionally, it was expected that there would be an increased use of an us(I) versus them dichotomy, contracting clauses, and heightened graduation. The analysis reveals an unexpected increase in negative judgements of the self and decreasing judgements of others, requiring further investigation. These findings also reveal a unique self-involved progression that gradually distances others through judgments of social sanction and contractive clauses. Finally, lexical infusion and repetition are found to enhance texts over time and contribute to an eventual heroic positioning. These results provide a linguistic perspective with the potential to aid in the detection of school shooters prior to the devastating events.

Poster Presentation Abstracts

(Alphabetical Order by Surname)

Sarah D’Antonio (Cornell University ~ United States)     

 How reasonable is the word 'reasonable'? A Corpus Study of Jury instructions

In the American Court System, a jury comprises a group of ordinary citizens who are typically not trained in law. A judge gives jurors instructions on their duties and certain relevant aspects of the law, which they then use to make a decision. Unfortunately, these instructions abound with linguistic complexities that impede understanding (Charrow & Charrow 1979; Tiersma 1999, inter alia). Thanks to efforts by Tiersma and others, many states have revised all or part of their instructions to increase comprehensibility. However, many problems exist even in the revisions. Previous studies on the well-studied instruction phrase “reasonable doubt” focus on the word “doubt”, which is indeed problematic (Solan 1999; Dhami et al. 2015, inter alia). Instead, though, this study contends with the vague non-intersective adjective “reasonable” — not only in “reasonable doubt”, but in other “reasonable” phrases within jury instructions: “reasonable person”, “reasonable care”, etc. Here, I will examine the jury instructions of Illinois, North Dakota, and Tennessee for each instance of a “reasonable” phrase. These three states have only partially revised their instructions: thus, each state’s revised and unrevised instructions can be compared. Assuming that the word “reasonable” is problematic, I hypothesize that the ratio of unrevised “reasonable” tokens to total “reasonable” tokens should be greater than the ratio of unrevised lines to total lines. Though this hypothesis is not borne out in a preliminary analysis of the Illinois criminal instructions — with ratios of 0.141 and 0.135 for overall lines, respectively — more data from the other states are needed.

Endurence Midinette Koumassol Dissake and Gratien Gualbert Atindogbé (University of Buea ~ Cameroon)
Interpreting in Cameroon courtrooms: Assessing the language proficiency of Interpreters

Cameroon is a multilingual nation with 284 languages (Simons and Fennig, 2017) amongst which two exoglossic languages, English and French. Official bilingualism, as it is termed, is the consequence of the colonial heritage of the country. Another aftermath of the colonial legacy of Cameroon is its bi-jural system, with the Common Law operating in the two English-speaking Regions, and the Civil Law in the eight French-speaking Regions. English as well as French-speaking Cameroonians can settle in any of the ten regions of the country. Yet, when people are involved in legal matters, they are judged in the language of the Region whether they are fluent in the latter or not. This use of language in the legal forum calls for a major concern: the assessment of language proficiency of the interpreters whose services are sought. The purpose of this paper is therefore to demonstrate the inadequacy (Hale, 2004, 2006 & 2010) of interpretation in Cameroon courtrooms. We abide to this statement of Moustacalis (in Todd 2008): “I am concerned that so many people who put their trust in the administration of justice … have suffered from incompetent interpretation. If you do not understand the proceedings through competent interpretation, you are denied justice”. As silent participant, we observed and assessed the competence of the interpreters in two court cases using the model of interpreting skills developed by Sandra Hale.

Sara Høyrup (Court Interpreter ~ Denmark)
The Interpreter as a Non-Expert

Problem: Most Danish court interpreters have no relevant university degree, in spite of regulations to the contrary, and most of the legal personnel in court ignore basic facts about professional translation. Legal staff often believes e.g. that the spoken translation must be “word by word” and fail to understand that interpreters need frequent breaks. Legal staff micro-manage the interpreters or even refrain from using them, they interrupt them and disbelieve their choice of wording if it is not literal enough. Legal staff instructs the interpreters to work in ways that do not function for the citizen, say by trying to speed up proceedings by ordering simultaneous interpretation for citizens that are in no state to handle the double stream of words. Legal staff clearly does not consider the court interpreters experts in their own craft. Consequence: Interpretation in Danish courts suffers from both the lacking expertise of many interpreters, and from the lack of respect surrounding the role as court interpreter. When the spoken translation is faulty, the citizen who does not master the trial language might misinterpret the questions and cannot express himself properly. He therefore does not get the fair trial that human rights entitle him to. Solution: training and certification of all court interpreters, improved remuneration and work conditions so as to attract and retain the best capacities. Training of the legal institutions’ personnel in the proper use of interpreters, including acknowledgement of the interpreters’ expertise in language and translation.

Ernst Kotzé (Nelson Mandela University ~ South Africa)
The Relative Value of Qualitative and Quantitative Data in Author Analysis: Reporting on a Case of Alleged Defamation

A constant problem of methodology in dealing with practical cases of author analysis and identification is the relative weight of qualitative and quantitative data in reaching a conclusion. In the case of quantitative data, statistical tests provide an acknowledged basis of comparison, and, very often, strong indicators towards shared or distinct authorship. However, a margin of error is always to be contended with, and a qualitative analysis is then required to provide a linguistic basis of comparison. This paper reports on a current case of alleged defamation after the break-up of a relationship which ended in court – incidentally, both parties are practising lawyers. Because of the strong emotional context, anonymous postings in social media were sufficiently prolific to provide sufficient data for analysis. A keyword analysis points to a strong possibility of common authorship of defamatory postings, but a stylistic analysis to investigate lexical and typographic similarities has been hampered by a deliberate attempt by the suspect to manipulate the text of the postings, e.g. by introducing spelling errors and grammatical deviations. The task of the forensic linguist in this case is therefore twofold: To establish whether different postings have a common author, and to establish a possible link between the suspect’s “normal” writing and that of the anonymous authors who produced the defamatory texts. The analysis of qualitative data in this case hence goes beyond a mere comparison of textual features – the possibility of double authorship within a single suspect has to be cleared.

Julien Plante-Hebert (Université de Montréal ~ Canada)
Forensic Voice Recognition of Familiar Speakers:Limitations and Novel Approaches

Humans still surpass machines when it comes to recognizing familiar voices. In our research, we aim to develop forensic applications of this human capacity. In a first study aiming to observe the effect of familiarity on speaker recognition using the voice lineup technique, results from 44 participants showed that recognition is particularly accurate for very familiar voices. Using an index of familiarity, accuracy rates surpassing 99.9% were obtained when the voice reflects that of a close friend, a relative, etc. However, the data shows that such accuracy only occurs with utterances of 4 syllables or more. Our study is one of the first to use voices of speakers who were personally known by the participants (instead of famous voices). In a legal context, however, there are limitations to the application of voice lineups in that some witnesses may not produce accurate answers in order to avoid incriminating a known individual. To answer this problem, we investigated the applicability of recording listeners’ electroencephalographic (EEG) responses to heard voices. In our second study, EEG responses were recorded from 11 listeners while they heard 4-syllable utterances produced by 14 different speakers of which only one was very familiar to the listener. The results showed a significant difference in ERPs for familiar voices. Overall, our experiment shows that it is possible under controlled conditions to demonstrate, beyond reasonable doubt, that a familiar speaker is recognized by an individual listener. Moreover, a standard protocol of voice lineup can be adapted for EEG recording.

Lauri Tavi (University of Eastern Finland ~ Finland)
Stress Detection Based on Acoustic Measurements of /i/-vowel

In this study, we investigate acoustic-phonetic parameters for the detection of psychological stress. We analyse speech of young adult females from eight authentic Finnish emergency call recordings. Based on the reason for the emergency call, the recordings are divided into two groups: the stress group and the neutral group. For stress detection, we use fundamental frequency (f0), the difference between the first and the second harmonics’ amplitude peaks (H1-H2), and centre of gravity. These acoustic features indicate, among other things, high speech tone and the location of energy of speech signal, which makes them theoretically justifiable stress detectors. We measure f0, H1-H2, and centre of gravity from manually segmented vowels and test stress detection accuracy with three different statistical classifiers: linear discriminant analysis, logistic regression, and decision tree models. Interestingly, our results show that all statistical models perform better when the models are trained using only vowel /i/ in comparison to the models that are trained with all the Finnish vowels /a/, /e/, /i/, /o/, /u/, /y/, /æ/, and /ø/. For instance, the use of our best performing model, a logistic regression based on only vowel /i/, yield 88 % correct classification, whereas logistic regression trained with all the vowels achieved the recognition rate of 80 %. We conclude that results from f0, H1-H2, and centre of gravity measures indicate good stress classification accuracy, yet further research with larger data sets is required.