Plagiarism involves intentional lifting of an idea and/or
intentional copying of the text (linguistic, musical, etc.)
used to express that idea, to cover up non-originality. So
it seems that sometimes it is the idea, sometimes it is the
language, sometimes it is both that are at stake; in other
words, that both form and content are relevant issues to its
definition.
In cases of plagiarism forensic linguists are called in to
help a Court decide whether two or more texts have been produced
independently or have been copied from an original text.
ForensicLab at IULA has specialised in the detection
of plagiarism in two specific areas:
- Plagiarism of an idea and /or of the text used to express
that idea in texts on the same topic.
- Plagiarism between translations.
The protocol used involves the following activities:
- To establish what markers and criteria are bound to be
taken as determinant in cases of plagiarism
- To answer a general empirical question: how similar two
or more texts have to be before we decide that one has been
plagiarized from another or others; how similar/how different
is enough?
- To establish the base-line one needs to have for degree
of similarity and the point at which this similarity becomes
suspicious.
- To indicate the rarity or otherwise of the choices made
by authors, and thus establish the uniqueness or commonality
of the texts produced. This is something that can only be
addressed through the use of corpora.
- To discuss how the other side, that is, the prosecution
or the defence, might respond to such evidence.
- To address another fundamental question: whether or not
this kind of evidence may help to establish a prima facie
case, that is, a case in which the evidence produced is
sufficient to enable a decision or verdict to be made unless
the evidence is rebutted.
The use of quantitative linguistic evidence can assist the
analyst in cases of plagiarism and add both internal and external
validity to the investigation. In order to determine the existence
of plagiarism we use the latest version of CopyCatch, called
CopyCatch Gold, designed by David Woolls from CFL Software
Development. This tool handles multiple document types and
allows rapid processing of the data in order to identify textual
similarity. The output is enhanced to show both phrase and
sentence matches and any text pair can be marked up to show
the way in which the shared vocabulary appears in the text.
The use of quantitative linguistic evidence (VARBRUL 2 y 3;
GOLDVARB 2001) can help the analyst in ensuring internal and
external validity in their investigation and in consultancy.
Measurements
- Overlapping vocabulary
- Shared once-only words: that is, words that
occur only once in each file (of the pairs compared) but
occur in both members of the pair. The higher the number,
both as a raw score and as a proportion of the shared vocabulary,
the greater the similarity between the two texts.
- Unique vocabulary: the number of words which do
not appear in the other text, in this case the words which
are unique to each translation. It would be assumed that
if two texts have been originally produced, they would include
a higher proportion of unique lexical items than those which
have not been produced independently.
- Shared once-only phrases: phrases that occur
only once in the two texts under comparison. The higher
the number, the lower the probability that the two texts
in the file comparison have been independently produced.
In cases of authorship determination/attribution, forensic
linguists are called in to help a Court decide among several
candidates whether the author of a particular set of texts
is also the author of another set of texts (authorship determinations)
, o whether one can attribute the authorship of a specific
text (for example, a suicide note, or extortion letters) to
a particular author (authorship attribution).
ForensicLab at IULA specialises in two specific areas
of authorship analysis:
- Authorship determination, where it is a question
of determining who, among several suspects from whom we
have a set of non-disputed texts, is the author of one or
several disputed texts, and demonstrating that these two
sets of texts (disputed and non-disputed) have not been
produced independently.
- Authorship attribution, where it is a question
of establishing if a disputed text or a set of disputed
texts can be attributed to a particular author, by contrasting
the disputed set of texts with a series of non-disputed
texts produced by this same author (control corpus).
The protocol used involves the following tasks:
- Establishment of the linguistic "markers" and the relevant
criteria in authorship determination.
- Indication of the rarity or expectancy of the choices
made by the authors so that the idiosyncrasy or coincidence
of the texts produced is established by means of using reference
corpora and data bases.
- Use of statistical tests to establish the significance
of the authorship marker values between different writers.
- Discussion on the possible counterattack by the other
part.
In order to determine and attribute authorship ForensicLab
uses reference corpora to establish the rarity/expectancy
of "linguistics markers" and marked and salient variables
by means of applying multivariate analysis (Varbrul 2 and
3; Goldvarb2001). Depending on the nature of the case (amount
of data, simple size, etc.), our laboratory can also use other
statistical techniques to establish that sets of disputed
ad non-disputed texts have been written by the same author.
At present, ForensicLab is developing an automatic tool and
technique of forensic authorship analysis that has been tested
in real and non real cases.
In cases of speaker identification forensic linguists are
called in to help a Court determine the authorship of a disputed
voice.
The Forensic Linguistics Lab at IULA specialises in
two specific areas:
- Voice-Speaker verification, in order to determine
if a disputed voice corresponds to a specific speaker.
- Speaker identification, in order to determine among
several candidates to which of them the disputed voice corresponds.
Voice-Speaker verification and Speaker identification
involve a joint analysis of voices from an auditive and acoustic
point of view. Auditive analysis does not use any equipment
for speaker identification and is rather based on the researcher's
experience and expertise. Acoustic analysis, on the
other hand, is based upon the analysis of the acoustic properties
of speech by means of speech technology (spectrogram,
voice multidimensional analysis, fundamental frequency, cochleagrams,
LTAS, etc.) and statistical analysis techniques.
When we listen somebody speak, the listener draws information
on the content (what is being said) and the person
speaking (voice quality). According to Rose (2002),
voice quality has two components: the organic component
and the articulatory component (o articulatory adjustment).
The former refers to the properties of sound which are determined
by the speaker's physiology (form and volume of the vocal
cavity, the specific characteristics of the vocal cords, etc.)
and the latter refers to the position and adjustment of the
articulatory organs (tongue, lips, etc.).

For forensic analysis of speaker identification all components
are important: in voice quality, the organic component is
determined by the physiological characteristics, and the articulatory
components by his/her idiosyncratic articulatory habits. Phonetic
quality is determined by dialect, sociolect, accent, verbal
habits, etc.
The protocol used in our laboratory involves the following
tasks:
- To determine what is the threshold level of intraspeaker
variation from which it will be possible to conclude that
a particular voice sample does not correspond to a particular
speaker's voice.
- To determine - both auditively and acoustically - which
voice parameters are more idiosyncratic of the speaker.
- To determine - both auditively and acoustically - the
specific characteristics of each speaker's dialect and sociolect
in order to count on arguments that would favour a positive
or negative identification.
- To compare acoustically the characteristics of several
sounds that appear in the disputed voice with those found
in samples of non-disputed texts and general reference corpora.
- To pursue a multivariate analysis of voice in order to
establish if two samples coincide in all the variables analyzed
(variables relating to fundamental frequency, amplitude
or bandwidth, frequency disturbance, noise levels, subharmonics,
etc.).
- To run statistical tests in order to establish the significant
values of the parameters under analysis. When no significant
differences are found, this indicates that the samples under
analysis are very likely to correspond to the same speaker.
In order to undertake acoustic analysis, the Forensic Linguistics
Laboratory uses the latest versions of the most reputable
programmes of acoustic analysis and voice edition, as well
as specific applications developed in our laboratory to engage
in comparative analysis. The acoustic data obtained are treated
statistically.
In cases of linguistic profiling forensic linguists are called
in to help in the construction of a suspect's profile. Drawing
from the linguistic analysis of the voice, one can establish
the author's geographical and social origin, gender, age,
weight, height, usual language, language contact, etc.
ForensicLab specialises in linguistic profiling in
Catalan, Spanish and English.
Linguistic profiling involves auditive and acoustic analysis
of unknown speakers' recordings in order to determine several
traits or characteristics that can be inferred from the voice:
origin, sociolect, gender and age, possible pathologies, weight
and height, usual language, language contact, etc.
The protocol used in our laboratory involves the following
tasks:
- To study the phonetic, phonological, morphological and
syntactic characteristics of the recording and to compare
them to the typical dialectal characteristics.
- To study the phonetic, phonological, morphological and
syntactic characteristics of the recording and to compare
them to the typical sociolectal characteristics.
- To undertake an analysis of the formants and fundamental
frequency from witch it is possible to infer weight and
height approximately.
- To identify linguistic interference with other languages
different from the language used in the recording.
In order to undertake acoustic analysis, the Forensic Linguistics
Laboratory uses the latest versions of the most reputable
programmes of acoustic analysis and voice edition, as well
as specific applications developed in our laboratory to engage
in comparative analysis. The acoustic data obtained are treated
statistically.
| TRANSCRIPTION
AND AUTHENTIFICATION OF RECORDINGS |
In cases of authentification of recordings, forensic linguists
are called in to clean out recordings from undesirable noise
and also to transcribe the recorded content. In cases where
the authenticity of a recording is questioned, Forensiclab
has at its disposal several tools to determine if such recording
has been manipulated.
| TRADEMARKS,
PATENTS AND FOOD LABELLING |
In cases of trademark and patent litigation, forensic linguists
are called in to make a linguistic analysis of patents and
to characterise trademarks by their appearance (orthography,
colour, typology and design), phonologically for sound and
semantically for meaning, in order to establish whether falsification
has occurred and confusion has been created. In food labelling,
forensic linguists can make proposals of new wordings for
ambiguous labels, which can cause a situation in which the
consumer fails to follow the guidelines provided.
| FORENSIC
LINGUISTIC MEDIATION |
On other occasions, forensic linguists are called in to analyse
the possible text ambiguity or text alteration in officialised
or private contexts or in legal, court and police written/spoken
discourse, involving either the administration or the users
themselves. In all these cases, forensic linguists analyse
the texts and mediate between parties.
- Dictamen pericial sobre un caso de Determinación de
Autoria, con sentencia a favor de la parte para la que se
actuaba.
SEHRS, multinacional turística catalana . October
2003 - March 2004
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Determinación de
Autoria, sin llegar a Vista Oral.
Rofes Asociados - Asesores en Derecho. January 2004
- May 2004
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Limpieza y Transcripción
de cintas audio (EC003155).
Private individual . January 2004 - March 2004
PRINCIPAL INVESTIGATOR: Dr. Jordi Cicres Bosch (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Determinación de
Autoria, sin llegar a Vista Oral (EC003166).
Planeta Agostini. January 2004 - May 2004
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección de Plagio,
con Vista Oral y sentencia favorable del Tribunal del Mercantil
para el que se actuaba como Perito (EC003180).
Magatzem Central de Llibreria. October 2004 - May
2005
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Marcas Registradas,
sin llegar a Vista Oral (EC003199).
Formas y Contenidos Jurídicos. S.L. January 2005 - March
2005
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Identificación
de Hablantes, sin llegar a Vista Oral (EC003206).
PSOE Asturias. January 2005 - May 2005
PRINCIPAL INVESTIGATOR: Dr. Jordi Cicres Bosch (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección de Plagio
entre libros de texto, con Vista Oral y sentencia favorable
para la parte por la que se actuaba (EC003214).
Grupo Editorial Bruño. January 2006 - June 2006
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Identificación
de Hablantes (EC003215).
Private individual. January 2006 - May 2006
PRINCIPAL INVESTIGATOR: Dr. Jordi Cicres Bosch (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección de Plagio
entre libros de texto (EC003221).
Private individual. November 2005 - June 2006
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Determinación de
Autoria, sin llegar a Vista Oral (EC003220).
Empresa CYBEX. February 2006 - October 2006
PRINCIPAL INVESTIGATORS: Dra. M. Teresa
Turell Julià (ForensicLab - IULA), Ms. Maria S. Spassova (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Interpretación
Semántica de Contratos, sin llegar a Vista Oral (EC003222).
Estudio Ghersi Abogados - Perú. September 2006 - November
2006
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección de Plagio
entre ASEREJÉ y RAPPER'S DELIGHT.
Sony ATV Music Publishing Holdings. Sociedad Cooperativa.
October 2007 - February 2008
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección de Plagio
entre libros de texto.
Private individual. November 2007 - April 2008
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Determinación/atribución
de Autoría, sin Vista Oral (EC003220).
Guardia Civil. November 2007 - February 2008
PRINCIPAL INVESTIGATORS: Dra. M. Teresa Turell (ForensicLab - IULA), Ms. Maria S. Spassova (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección
de Plagio entre libros de texto (EC003221).
Private individual. November 2005 - June 2006
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Determinación
de Autoria, sin llegar a Vista Oral (EC003220).
Empresa CYBEX. February 2006 - October 2006
PRINCIPAL INVESTIGATORS: Dra. M. Teresa Turell
Julià (ForensicLab - IULA), Ms. Maria S. Spassova (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Interpretación
Semántica de Contratos, sin llegar a Vista Oral (EC003222).
Estudio Ghersi Abogados - Perú. September 2006 - November
2006
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección
de Plagio (National Geographic). Pendiente de Vista
Oral (EC003218).
Private individual vs. RBA Publicaciones. May 2006
- February 2007
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección de Plagio
entre ASEREJÉ y RAPPER'S DELIGHT.
Sony ATV Music Publishing Holdings. Sociedad Cooperativa.
October 2007 - February 2008
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Detección
de Plagio entre libros de texto.
Private individual. November 2007 - April 2008
PRINCIPAL INVESTIGATOR: Dra. M. Teresa Turell
Julià (ForensicLab - IULA)
- Dictamen pericial sobre un caso de Determinación/atribución
de Autoría, sin Vista Oral (EC003220).
Guardia Civil. November 2007 - February 2008
PRINCIPAL INVESTIGATORS: Dra. M. Teresa Turell (ForensicLab - IULA), Ms. Maria S. Spassova (ForensicLab - IULA)