Assessing the veracity

of children's recantation
of sexual abuse
disclosures
Presentation prepared by Olumide Popoola

Germanic Society of Forensic Linguistics Roundtable 2015
Case Background

- Brother and sister aged 8 and 9 made allegations of sexual abuse with
ritualistic elements involving their father and teachers at their school.
- Disclosure first made to mother and her new partner ('Abraham') and then to
the local police.
- Police conducted three Achieving Best Evidence (ABE) interviews with the
children. In the first two interviews the children maintained their
allegations.
- In the third interview the children told a different story and blamed 'Abraham'
for 'making them say' what they had said before.
- The criminal investigation was closed and the case deemed to be an issue for
the Family Courts.
Recantation and retraction in child sexual abuse cases
" Investigating child sexual abuse allegations, Malloy et al. (2007) found that
children’s vulnerability to adult familial influences predicted recantation
(i.e., those who were younger, had made accusations against a parent
figure, and had nonoffending caregivers who reacted unsupportively to
disclosure were more likely to recant). However, recantation was unrelated
to whether children’s allegations were corroborated by external evidence
(e.g., medical evidence, suspect admission) or to whether custody issues
affected the involved parties. In other words, it appears that at least some
children recanted true allegations of sexual abuse, seemingly due to familial
pressures. It is imperative to be attentive to both concerns about the risk of
false allegations as a result of children’s suggestibility, and the potential
external pressures that may exist for children (and adults) to falsely deny or
recant allegations of wrongdoing."
from 'Recantation in Legal Contexts' By Lindsay C. Malloy, Jillian Rivard, MA,
Allison P. Mungo, BA, and Peter Molinaro, MA AP-LS News | March 2014
Hypothesis (1)
Based on Crown Prosecution Service guidelines and and police interview best
practice as recommended in the Achieving Best Evidence guidelines, a bona
fide retraction will
1) contain moves that:
- ask for and give reasons for changing the story
- ask for and give reasons for lying in the first place
- will have a preponderance of open questions
- will not have suggestive or leading questions
Move analysis can assess this. The Swalesian approach to move analysis is
notoriously subjective. However, William Mann's Rhetorical Structure
Theory (RST) and Dialogue Macrogame Theory (RST) have a set of moves
that have been validated across a number of genres and and text types plus
the ability to incorporate (invent) new moves into the framework.
Comparison of section in A and G interviews discussing the disclosure
that “Papa kills babies” indicates significantly different rhetorical structure.
Interview A Interview G

Elaboration 9 Verify 8
Reason 8 Elaboration 3
Summary 3 Suggested 2
Alternative
Interpretation 2
Concession 2
Sequence 2
Circumstance 2
Purpose 1
Reason 1
Circumstance 1
Consequence 1
Background 1
Restatement 1
Solution 1
Concession 1
Interpretation of “Papa kills babies” episodes in
Interview A
In Interview A, the police interviewer invites A to give reasons for the original
disclosure that “Papa kills babies” through open-ended questions that give A the
opportunity to elaborate. A is keen both to explain how Abraham (her mother's new
partner) pressured her to make the disclosure about her father and to tell the story
of 'Mask of Zorro', a film she clearly enjoyed and remembers well. In two
Interpretation moves, A voluntarily makes a comparison between the “bad man
from California” and her father.
The police interviewer prompts A to talk about the use of a knife in the film since she
had mentioned this before both in relation to the film discussion prior to this
interview and in her original disclosure alleging that her father cuts babies heads.
Eventually A narrates that the 'baddie who looks like her Dad' cuts a man's head
and hands off in the film. The police interviewer makes explicit the implicit that A's
disclosure that “Papa kills babies” comes from the events of 'Mask of Zorro' with
the Summary move:

“Ok, so this is- so that's the man who looks like your Dad in the film who done that.
/OK\.”
Interpretation of “Papa kills babies” episodes in
Interview G
In this interview, G maintains that his father has killed “some babies” but concedes
that the number killed is not as many as stated in previous disclosures (“not
much”). G maintain this position despite being pressured by the police interviewer
to change his story through repeated verification moves and a suggested
alternative.
Eventually G does change his story and states that he had lied in his previous
disclosure. Whether this is an admission or a response to the suggestions of the
police interviewer cannot be definitively determined here. However, it is clear that
the police interview deploys moves and a questioning style that is not consistent
with ABE interviews.
Hypothesis #2
For a recantation to be given the legal status of a retraction (and therefore end
an investigation) the statement must be unequivocal and unambiguous,
credible as well as truthful.
The concept of veracity, recently developed as a dimension of information
quality (IQ) for use in big data evaluation for business and for library
information management systems, will be a useful way of assessing the
evidentiary value of a victim (or witness) recantation
Lukioianova and Rubin (2013) provides a comprehensive theoretical framework
for and ways of measuring veracity for big data starting from the position
that “veracity goes hand in hand with inherent uncertainty” . Veracity exists
on a certainty-uncertainty dimension; deception is the opposite of truth and
is taken as one component of veracity (along with objectivity and
credibility).
Following this, I hypothesized that a bona fide retraction will display high levels
of certainty and low levels of deception.
Examples of Certainty Markers used in Interviews A and G

Absolute Certainty EVERYTHING ; ALL ; NEVER EVEN; ANYTHING.THE
WHOLE THING; IT'S ONLY; STRAIGHTAWAY; EVERY
SINGLE DAY; I DON'T CARE

High Certainty TOO SCARED; ACTUALLY; THAT'S WRONG; A WHOLE BIG
PACKET; KEPT ON ARGUING; KNOWS ABOUT IT; LOADS
OF HOUSES; REALLY HOPING; THEY DID FIND;

Moderate Certainty SOME OF THEM; WE GUESSED; FUN BUT NOT SO MUCH
FUN.; WOULD BE; THINK; WAS PLANNING; NOT REALLY;
YOU PROMISED ME ; JUST IN CASE

Low Certainty MIGHT;IF IT'S TRUE OR NOT; HOW'S IT CALLED;
HOPING;

Uncertainty YEAH THERE IS THOSE ROOMS. NO THERE ISN'T ; HAVE
GONE.BLANK; DON'T KNOW.; NO IDEA
As the hypothesis suggested, both interviews contain high levels of
certainty.
CERTAINTY Interview A – Interview A - % Interview G – Interview G -
MARKER LEVEL Total Total %

Absolute 27 13% 13 12%

High 28 14% 14 13%
Moderate 11 5% 8 8%

Low 4 2% 2 2%
Uncertain 10 5% 3 3%

Unmarked 121 60% 66 62%

Total sentences 201 106
However, contradictory statements indicate that high
certainty language is being used for deception
INTERVIEW A INTERVIEW G
63 Police: Ok. And so what about all your teachers 108 Police: oh ok. so does your dad live in
and all that= sheffield?

64 A. =THEY DON'T DO IT. 109 G: YEAH BUT HE STAYS IN HIS OTHER
FRIEND'S HOUSE THAT I DON'T KNOW AND
65 Police: They don't do it. HE STAYS AT MR HOLLINS' HOUSE AND SOME
DAYS HE GOES OFF BACK TO SHEFFIELD.
66 A. THEY NEVER EVEN MET MY.. MR HOLLINGS
HE'S A TEACHER, HE NEVER EVEN M- MET MY 110 Police: Ok and do you know he stays at Mr
DAD Hollins' house or do you guess that?

67 A: HE DOESN'T EVEN KNOW HIS NAME. 111 G: WHO? NO, NO NO I DON'T.

68 Police: Have you been to his house= 112 Police: You don't know, you guess or

69. A: NO 113 Q: NO I DON'T GUESS BECAUSE UM
BECAUSE MY DAD AND MR HOLLINS ARE
REALLY CLOSE FRIENDS, REALLY CLOSE.
The interviews were successfully segmented into episodes on the basis of
topics of questioning. Although overall levels of certainty were similar for
the two interviews, the variance was significantly higher for Interview A,
meaning that interviewee A is lurching from certain to uncertain i.e. has
wildly inconsistent and changeable language.
A connection between language change and deception has been identified in
Picornell (2013 and 2014), albeit on a broader set of linguistic cues that
includes pronouns as well as 'ambiguity markers'. However, based on this
data from two interviews, I would hypothesize that the high inter-episode
variance of Interview A is more consistent with deception than the
comparatively constant distribution of certainty markers in Interview G.
NTERVIEW INTERVIEW
G

pisode Total number Number of % Certainty Episode Total Number of % Certa
of sentences Certainty Marked number of Certainty Marked
Marked sentences sentences Marked sentenc
sentences sentences

1 6 1 17% 1 6 2 3

2 17 5 29% 2 11 5 4

3 32 10 31% 3 12 4 3

4 11 3 27% 4 7 2 2

5 20 5 25% 5 4 1 2

6 13 6 46% 6 6 2 3

7 10 1 10% 7 9 4 4

8 18 11 61% 8 12 4 3

9 12 3 25% 9 15 8 5

10 26 14 54% 10 8 3 3

11 9 6 67% 11 16 5 3

12 20 12 60%

13 7 5 71%
The box plot analysis revealed an outlier episode in Interview G; Episode 9 was where the police
interviewer directly contradicted the assertions of sexual activity made by G by making the
knowledge claim that G's sister had already said that opposite was true. The following language
from G contains an uncharacteristically high level certainty markers
INTERVIEW A INTERVIEW G

Episode 9

Median: 31
Interquartile Median: 33
range: 33
Interquartile
Population size:13 Range: 13
Lower quartile: 25
Upper quartile: 58 Population size:11
Lower quartile: 31
Upper quartile: 44

Levene's Test for Homogeneity of
Variance = p = 0.02128

Wilcoxon rank-sum test
p = .9768
Episode 9 124 P: Why was you nervous about telling me that none of
this was real.
114 Police: OK. And the, the stuff with the plastic
willies, how did you find out about the plastic 125 G: BECAUSE ER ABRAHAM ALREADY GOT YOU...HE
willies, who told you about plastic willies ALREADY GOT YOU OF IT...AND HE ER YOU KNOW THE
FIRST TIME HE EVER FOUND OUT ME AND MY SISTER
115 G: NO NO THAT'S TRUE. TOUCHING????, IT WASN'T ACTUALLY TRUE WE ARE
KISSED US...WE ACTUALLY NEVER TOUCHED, I KNOW
116 Police: That's true? IT'S AGAINST THE LAW AND WE NEVER TOUCHED,

117 G: YEAH 126 P: never touched

118 P: OK. Coz I've been told something different. 127 G: NO WE NEVER TOUCHED EACH OTHER.

119 G:................BY[G's sister name]? 128 P: OK. You've been to the doctor's though haven't you.

120 P: Yeah......so honestly, just be truthful...if things 129 G: YES
aren't..I promise you won't get in trouble if 130 P: And they said, there are, did they tell you what
you've lied before, I promise. But we just, all it is they found, the doctors, what did they say?
I need to know the truth, what to look for and
who to speak to. So if some of it is a lie, and 131 G: BRUISES
you've lied about something I promise you won't
132 P: Bruises ok. And where were the bruises?
get into trouble. OK?
133 G: ON MY BOTTOM.
121 G: SO I LIED ABOUT THE PLASTIC WILLIES.
134 P: And how did you get them?
122 P: How did you find about plastic willies?
135 Q: I GOT NO IDEA.
123 G: NONE OF IT WAS REAL. NONE OF IT.
Discussion
Hypothesis 1: The rhetorical structure of the two interview excerpts shown is clearly different
and the move analysis serves to highlight that fact. Interview A looks like a retraction in that
the questions are open and reasons for prior disclosure and current recantation are
requested and given. This is in stark contrast to Interview G which contains suggestion and
challenging questioning more suitable for obtaining a confession than allowing a voluntary
recantation.

Hypothesis 2: Contrary to my initial hypothesis, the interviews contain both high certainty
markers and indications of deception. This may be due to the interviewee attempts to sound
convincing. Deception is explicitly shown in the contradictory statements across Interviews A
and G regarding the relationship between Mr Hollings and the children's father. The extent of
language variance in A's interview may suggest that this is the more deceptive interview of
the two, despite it appearing to follow the correct format for a retraction.

Conclusion: The analysis has highlighted the interviews' lack of structural integrity and presence
of uncertainty to the extent that it would be a considerable leap of faith to give them the
status of a retraction and use this to close an investigation. The method has produced some
interesting possibilities for further research into 1) the analysis of deception in non-narrative
witness statements, 2) the correlation between statistically significant language variance
and deception and 3)the investigative power of identifying statistical outliers.
Selected References

Lyon, T. D. (2014). Interviewing children. Annual Review of Law and Social Science, 10, 73-89.

Malloy, L. C., Lyon, T. D., & Quas, J. A. (2007). Filial dependency and recantation of child sexual
abuse allegations. Journal of the American Academy of Child & Adolescent Psychiatry, 46(2),
162-170.

Picornell, Isabel. "Analysing deception in written witness statements." Linguistic Evidence in
Security, Law and Intelligence 1.1 (2013): 41-50.

Rubin, V. L., & Vashchilko, T. (2012). Extending information quality assessment methodology: A
new veracity/deception dimension and its measures. Proceedings of the American Society
for Information Science and Technology, 49(1), 1-6

Rubin, V. L., & Lukoianova, T. (2015). Truth and deception at the rhetorical structure level.
Journal of the Association for Information Science and Technology, 66(5), 905-917.

Rubin, V. L. (2010). Epistemic modality: From uncertainty to certainty in the context of
information seeking as interactions with texts. Information Processing & Management,
46(5), 533-540.