Can the Words People Use to Answer “Ordinary” Questions Reveal Childhood Sexual Abuse?

This paper is based on Ginger N. Smith (Northwest Vista College), Sember A. Lucas (Texas Tech University), and Donald R. Lucas’ (Northwest Vista College) 2023 Southwestern Psychological Association convention talk titled, Can the Words People Use to Answer “Ordinary” Questions Reveal Childhood Sexual Abuse?

Let’s begin with some numbers — some extraordinary yet true numbers.

About four million people will be born in the United States this year.

Of these four million people, more than 10% of them — that is 400,000 of them, will be sexually abused before their 18th birthday.

And of these 400,000 Americans, less than 20% of them will report the sexual abuse.

Not surprisingly childhood sexual abuse is traumatic and has significant potential short-term and long-term consequences.

Some of the short-term and long-term consequences of childhood sexual abuse.

These consequences can be mediated. But are most likely to be mediated — only after the childhood sexual abuse is reported. For a complete list of mediators of childhood sexual abuse, see Smith, Lucas, and Lucas 2022b.

Some of the mediators of childhood sexual abuse.

HYPOTHESIS

This brings us to the basic question driving the present study: Is it possible to increase the number of people reporting their childhood sexual abuse?

If this is possible, then more people — who have been sexually abused in their childhoods, will be able to mediate the potential consequences of their abuse.

The present study explored this basic question by testing if it is possible for a computer-based text analysis program to determine whether a person was sexually abused in childhood.

We did this by using a sample of participants in which we knew were sexually abused in their childhood.

METHODS

We had community college students complete a 106-question digital survey. For a comprehensive description of the survey, see Smith, Lucas, and Lucas 2022b.

The survey was composed of six types of questions: Demographic, Lie, Reliability, Childhood Sexual Abuse, Consequence, and Mediator questions.

Of the 406 community college students who completed the survey, 346 of them passed the Lie and Reliability tests.

And of these 346 students, 117 were determined to have been sexually abused in their childhoods, whereas 229 were determined to not have been sexually abused in their childhoods.

We used seven of the 106 questions on the survey to determine if a student was sexually abused in childhood (see Figure 4).

Figure 4: Survey questions used to determine childhood sexual abuse.

If there was an affirmative answer to one or more of these seven questions, then the student was determined as being sexually abused in childhood. For example, if a student answered yes to the question, “Between birth and 17 years of age, did you ever have a sexual experience with someone five years or older than you?” Then they were determined as being sexually abused in childhood. Or, if a student answered yes to the question, “Between birth and 17 years of age, did you ever have a sexual experience with someone you considered a caregiver?” Then they were determined as being sexually abused in childhood.

We analyzed the answers students — who were sexually abused in childhood, as well as those students who were not sexually abused in childhood, gave to five “open-ended” questions.

These questions are:

Write three words that best describe your home environment from your birth to 17 years of age…

Write three words that best describe what a man is…

Write three words that best describe what a woman is…

Write three words that best describe your romantic relationships…

Write three words that describe your definition of sexuality…

Note, these questions are relatively “ordinary.” None of these questions are about abuse or trauma and only one of the questions is about sexuality.

We analyzed students’ answers using Linguistic Inquiry and Word Count –version 22 or LIWC-22.

Created in 1993 LIWC is a scientifically-validated computer-based text analysis program that has been used in more than 20,000 research articles.

LIWC analyzes 110 dimensions of language.

We used ten of LIWC’s dimensions to analyze students’ answers. These dimensions are Affect, Positive Tone, Negative Tone, Emotion, Positive Emotion, Negative Emotion, Social Processes, Social Behavior, Prosocial, and Sexual.

LIWC calculates these dimensions relative to a dictionary of words for the dimension. The more words a person says that match the words within LIWC’s dimension’s dictionary, the higher that person’s dimension score.

For example, the words “good, well, new, love” are four of the 1,020 words within LIWC’s dimension’s dictionary for Positive Tone. If a participant in our study answered a question with the words, “bad, hate, love,” then the participant would have a Positive Tone dimension score of 33%. One of the three words the participant said (that is, 33%) was in the Positive Tone dictionary. This same answer would yield a score of 66% on the Negative Tone dimension because two of the three words the participant used in their answer are in LIWC’s dimension’s dictionary for Negative Tone. If a study participant answered a question with words not in a dimension’s dictionary, then the participant’s score on the dimension the dictionary is associated with would simply be 0.

RESULTS

Write three words that best describe your home environment from your birth to 17 years of age…

The 117 participants who were sexually abused in childhood (hereafter referred to as CSU participants) gave 57 different answers. Whereas the 229 participants who were not sexually abused in childhood (hereafter referred to as NoCSU participants) gave 72 different answers. (A complete list of participants’ answers is available upon request.) The top three most frequent answers participants gave differed between groups (see Figure 5). Both groups of participants gave LOVING as their most frequent response, but then CSU participants said CHAOTIC and TOXIC as their next most frequent answers, whereas the NoCSU participants followed LOVING with the answers SAFE and FUN.

Figure 5: Most frequent answers to the question, “Write three words that best describe your home environment from your birth to 17 years of age…”

LIWC found CSA participants as compared to NoCSA participants gave statistically significant less Positive Tone (24%, 42%, respectively), Positive Emotion (15%, 25%, respectively), and Prosocial (10%, 15%, respectively) words in their answers (ps < .05).

Write three words that best describe what a man is…

The 117 CSA participants gave 48 different answers. Whereas the 229 NoCSA participants gave 85 different answers. (A complete list of participants’ answers is available upon request.) The top three most frequent answers participants gave did not differ between groups (see Figure 6). Both groups of participants gave STRONG as their most frequent response followed by CARING and PROTECTOR.

Figure 6: Most frequent answers to the question, “Write three words that best describe what a man is…”

LIWC found CSA participants as compared to NoCSA participants gave statistically significant less Positive Tone (31%, 41%, respectively) and Prosocial (10%, 16%, respectively) words in their answers (ps. < .05)

Write three words that best describe what a woman is…

The 117 CSA participants gave 40 different answers. Whereas the 229 NoCSA participants gave 65 different answers. (A complete list of participants’ answers is available upon request.) The top three most frequent answers did not differ between groups (see Figure 7). Both groups of participants gave STRONG as their most frequent response followed by CARING and INDEPENDENT.

Figure 7: Most frequent answers to the question, “Write three words that best describe what a woman is…”

LIWC found CSA participants as compared to NoCSA participants gave statistically significant less Positive Emotion (7%, 9%, respectively) words in their answers (p. < .05).

Write three words that best describe your romantic relationships…

The 117 CSA participants gave 57 different answers. Whereas the 229 NoCSA participants gave 75 different answers. (A complete list of participants’ answers is available upon request.) The top three most frequent answers substantially differed between groups (see Figure 8). Both groups of participants gave LOVING and FUN within their top three, but the third most frequent answer for the CSA participants was TOXIC, whereas the third most frequent answer for the NoCSA participants was HAPPY.

Figure 8: Most frequent answers to the question, “Write three words that best describe your romantic relationships…”

LIWC found CSA participants as compared to NoCSA participants gave statistically significant more Negative Tone (20%, 8%, respectively) words in their answers (p < .001) and statistically significant less Social Behavior (19%, 24%, respectively) and Social Processes (20%, 26%, respectively) words in their answers (ps < .04).

Write three words that best describe your definition of sexuality…

The 117 CSA participants gave 50 different answers. Whereas the 229 NoCSA participants gave 90 different answers. (A complete list of participants’ answers is available upon request.) The top three most frequent answers slightly differed between groups (see Figure 9). Both groups of participants gave OPEN within their top three, but the CSA participants included ATTRACTION and FREE, whereas the NoCSA participants included LOVING and FLUID.

Figure 9: Most frequent answers to the question, “Write three words that best describe your definition of sexuality…”

LIWC found none of the ten language dimensions differ between the answers given by the CSA and the NoCSA participants.

DISCUSSION

With 5,000 to 25,000 children for every 100,000 being sexually abused, there is no other way to say it: Sexual abuse is a childhood epidemic. For comparison, less than 20 out of every 100,000 children will get cancer.

Despite these epidemic numbers and for a variety of reasons, only a fraction of sexually abused children report it.

Without reporting and without awareness, childhood sexual abuse is much more likely to have long-term consequences.

The present study found analyzing the psychological value of words people use to answer ordinary questions can reveal childhood sexual abuse and thereby may be used as a tool to increase the number of people reporting their abuse.

Specifically, the present study found people who were sexually abused in childhood use significantly different positive tone, negative tone, positive emotion, prosocial, social behavior, and social processes in the words they use to answer ordinary questions.

Generally, those who were sexually abused in childhood tend to use less positive value words than those who were not sexually abused in childhood. To be clear on this point, the present study did not generally find those who were sexually abused in childhood to be more negative in their tone or emotions but instead just less positive.

The present results are evidence for psychoanalytic theories about childhood sexual trauma being reflected in language usage — especially when the language is not intentionally monitored by the speaker. Indeed, we never asked our participants directly about their sexual abuse. The closest we came was when we asked them to define sexuality — which was the only question whose answers did not significantly differ between those who were sexually abused in childhood to those who were not sexually abused in childhood.

These results must be viewed relative to several limitations in this study.

The first being, the limited number of dimensions we used to analyze participants’ answers — future research should take greater advantage of the more than 100 dimensions LIWC allows words to be analyzed by.

The second limitation is how we gathered the participants to be a part of this study. Our sample was a sample of convenience relative to the professors who answered our emails and allowed their students to participate.

The third limitation is this is a retrospective correlational study — although we found statistically significant differences between how those who were sexually abused in childhood answered questions and those who were not sexually abused in childhood answered questions, the present study design does not allow us to say these differences were due to the childhood sexual abuse.

Lastly, this is the first study of its kind, thus the present results have yet to be replicated.

Despite the limitations of this study, imagine three things:

First, imagine a clinician in a therapy session with a client — a client the clinician suspects was sexually abused in childhood. The therapy session is guided by simple, non-triggering questions. The answers the client gives to these questions are recorded, then analyzed by a computer-based text analysis program which probabilistically determines if the client was sexually abused as a child.

Next imagine a client suspected of being sexually abused in childhood being prescribed by their clinician, a computer program. The client — on their own, sits with the program. The program poses ordinary, non-triggering questions to the client for them to answer. The computer program then compares the client’s answers to the answers of those known to have been sexually abused in childhood and returns to the client a probability value of whether they were sexually abused in childhood.

Lastly, imagine the client is a child sitting with themselves in front of a clinician-prescribed computer program. The child’s clinician suspects the child is being sexually abused. The computer program poses ordinary, non-triggering questions for the child to answer. The computer program then compares the child-client’s answers to the answers of those children known to have been sexually abused in their childhoods. The computer program returns to the clinician a probability value of whether their child-client is being sexually abused.

All this imagining is for one purpose: To objectively address, mediate, and end the biggest epidemic facing children today — sexual abuse.

REFERENCES

Beitchman, J. H., Zucker, K. J., Hood, J. E., DaCosta, G. A., Akman, D., & Cassavia, E. (1992). A Review of the Long-Term Effects of Child Sexual Abuse. Child Abuse & Neglect, 16(1), 101–118.

Boyd, R. L., Ashokkumar, A., Seraj, S., & Pennebaker, J. W. (2022). The Development and Psychometric Properties of LIWC-22. Austin, TX: University of Texas at Austin.

Briere, J. N., & Elliott, D. M. (1994). Immediate and Long-Term Impacts of Child Sexual Abuse. The Future of Children, 4, 54–69.

De Bellis, M. D., & Zisk, A. (2014). The Biological Effects of Childhood Trauma. Child and Adolescent Psychiatric Clinics of North America, 23, 185–222.

Gewirtz-Meydan, A., & Finkelhor, D. (2020). Sexual Abuse and Assault in a Large National Sample of Children and Adolescents. Child Maltreatment, 25, 203–214.

Hamilton, B.E., Martin, J.A., Osterman, M.J.K. Births: Provisional Data for 2021. Vital Statistics Rapid Release; no 20. Hyattsville, MD: National Center for Health Statistics. May 2022.

London, K., Bruck, M., Ceci, S. J., & Shuman, D. W. (2005). Disclosure of Child Sexual Abuse: What Does the Research Tell Us About the Ways That Children Tell? Psychology, Public Policy, and Law, 11, 194–226.

National Center for the Victims of Crime. Child Sexual Abuse Statistics.

Pennebaker, J. W., & Stone, L. D. (2003). Words of Wisdom: Language Use Over the Life Span. Journal of Personality and Social Psychology, 85, 291–301.

Smith, G., Lucas, S. A., & Lucas, D. R. (2023). Can the Words People Use to Answer “Ordinary” Questions Reveal Childhood Sexual Abuse? Paper presented at the annual meeting of the Southwestern Psychological Association, Frisco, Texas.

Smith, G., Lucas, S. A., & Lucas, D. R. (2022a). Incidence, Consequences, and Mediators of Childhood Sexual Abuse in Community College Students. Paper presented at the annual meeting of the Southwestern Psychological Association, Baton Rouge, Louisiana.

Smith, G., Lucas, S. A., & Lucas, D. R. (2022b). Incidence, Consequences, and Mediators of Childhood Sexual Abuse in Community College Students. Medium, April 26.

United States Department of Justice — Justice Programs. Sexual Abuse in America: Epidemic of the 21st Century.

United States Environmental Protection Agency. Health — Childhood Cancer.

Zwickl, S., & Merriman, G. (2011). The Association Between Childhood Sexual Abuse and Adult Female Sexual Difficulties. Sexual & Relationship Therapy, 26, 16–32.

…

Ginger Smith is a sophomore-level psychology major at Northwest Vista College. Sember Lucas is a senior-level psychology major at Texas Tech University. Dr. Don Lucas, Ph.D. is a Professor of Psychology and head of the Psychology Department at Northwest Vista College in San Antonio Texas.

If you are interested in the science of human sexuality, then check out https://humansexuality.medium.com/ and https://www.youtube.com/channel/UCQFQ0vPPNPS-LYhlbKOzpFw/

science, social sciences, childhood sexual abuse, childhood trauma, clinical psychology, child abuse, human sexuality, sex education, health psychology, community college education, language, development