Мир закона

МРНТИ 10.79.00

УДК 343.98:159.964.2

Schafer Jack — Professor, Law Enforcement and Justice Administration, Western Illinois University, PhD (Macomb IL, USA);

Stinson Troy — Professor, Law Enforcement and Justice Administration, Western Illinois University, PhD (Macomb IL, USA);

Meloni Thomas — Professor, Law Enforcement and Justice Administration, Western Illinois University, PhD (Macomb IL, USA)

DETECTING DECEPTION IN WITNESS STATEMENTS

Annotation. This study examined the predictive value of grammar structures to differentiate truthful witness statements from deceptive witness statements. Participants were instructed to write false witness narratives implicating themselves in a crime they did not commit. The five criteria examined were text bridges, tense changes, spontaneous negations, word qualifiers, and the word «just». The study results showed that the combination of three variables, text bridges, spontaneous negations, and tense changes, correctly predicted deceptively written narratives 80 percent of the time and matched the precision of the Incident Driven Deception Detection (ID-3) computer program at 90 percent. This study proposes a new approach to assess the veracity of written narratives.

Keywords: deception detection; witness statements; text bridges; transitional word, spontaneous negations; tense changes; verbal cues; false confessions, grammar structures.

Шафер Джек — Батыс Иллинойс университетінің құқық қорғау және әділет басқармасы, PhD, профессор (Макомб, Иллинойс, АҚШ);

Стинсон Трой — құқық қорғау және сот төрелігін басқару, Батыс Иллинойс университеті, PhD, профессор (Макомб, Иллинойс, АҚШ);

Мелони Томас — құқық қорғау және сот төрелігін басқару, Батыс Иллинойс университеті, PhD, профессор (Макомб, Иллинойс, АҚШ)

КУӘГЕРЛЕРДІҢ АЙҒАҚТАРЫНДА АЛДАУДЫ АНЫҚТАУ

Түйін. Бұл зерттеу шынайы куәгерлердің айғақтарын жалған куәгерлердің айғақтарынан ажырату үшін грамматикалық құрылымдардың болжамды мәнін зерттеді. Қатысушыларға жалған айғақтар жазу тапсырылды, онда олар өздерін жасамаған қылмысы үшін айыптады. Қарастырылған бес критерий: мәтіндік көпірлер, уақыттың өзгеруі, өздігінен бас тарту, сөз детерминанттары және «жай» сөзі. Зерттеу нәтижелері үш айнымалының, мәтіндік көпірлердің, стихиялық теріске шығарулардың және уақыттың өзгеруінің үйлесімі уақыттың 80 пайызында алдамшы жазылған әңгімелерді дұрыс болжағанын және 90 пайыз Incident Driven Deception detection (id-3) компьютерлік бағдарламасының дәлдігіне сәйкес келетінін көрсетті. Бұл зерттеу жазбаша әңгімелердің дұрыстығын бағалаудың жаңа әдісін ұсынады.

Түйінді сөздер: алдауды анықтау, куәгерлердің айғақтары, мәтіндік көпірлер, өтпелі сөз, өздігінен бас тарту, уақыттың өзгеруі, ауызша сигналдар, грамматикалық құрылымдар.

Шафер Джек — Управление правоохранительными органами и правосудием, Университет Западного Иллинойса, PhD, профессор (США, Макомб, Иллинойс);

Стинсон Трой — Управление правоохранительными органами и правосу­дием, Университет Западного Иллинойса, PhD, профессор (США, Макомб, Иллинойс);

Мелони Томас — Управление правоохранительными органами и правосу­дием, Университет Западного Иллинойса, PhD, профессор (США, Макомб, Иллинойс)

ОБНАРУЖЕНИЕ ОБМАНА В ПОКАЗАНИЯХ СВИДЕТЕЛЕЙ

Аннотация. В этом исследовании изучалась прогностическая ценность грам­матических структур для отличия правдивых показаний свидетелей от ложных по­ка­заний свидетелей. Участникам было поручено написать ложные свидетельские рассказы, в которых они обвиняли себя в преступлении, которого не совершали. Пять рассмотренных критериев: текстовые мосты, изменения времени, спонтан­ные отрицания, определители слов и слово «just». Результаты исследования пока­зали, что сочетание трех переменных, текстовых мостов, спонтанных отрицаний и изменений времени, правильно предсказывало обманчиво написанные рассказы в 80 процентах случаев и соответствовало точности компьютерной программы Incident Driven Deception Detection (ID-3) в 90 процентах. В этом исследовании предлагается новый подход к оценке правдивости письменных рассказов.

Ключевые слова: обнаружение обмана, показания свидетелей, текстовые мосты, переходное слово, спонтанные отрицания, смена времени, вербальные сигналы, грамматические структуры.

 

Introduction. Law enforcement officers have increasingly relied on narrative analysis to assess the veracity of written narratives. However, empirical research has yet to identify specific linguistic cues that portend deception. Detectives, federal agents, insurance fraud investigators, and others have used various methods to analyze statements of suspects and alleged victims. These methods include: Investigative Discourse Analysis (Rabon, 1996)[1], Scientific Content Analysis (Sapir, 1987)[2], Statement Analysis (Kaster, 1999)[3], Verbal Behavioral Analysis (Rudacelle, 1994)[4], Criteria-Based Content Analysis (Steller & Koehnken, 1989)[5], and Reality Monitoring (Johnson and Raye, 1981)[6]. These analyses have identified various linguistic cues that may signal deception in written communications; however, the reliability of the linguistic cues across narratives has yet to be demonstrated. This study analyzed the linguistic characteristics of witness statements and suggests a new method to evaluate the truthfulness of written accounts.

Witness Narratives

A written witness statement is defined as a narrative written by an individual, without external influence, about an event they witnessed or experienced. Witnesses’ accounts of events are subjective and shaped by their emotions and social experiences. Witnesses often reconstruct events to match their perspective, including rearranging or omitting details. Additionally, word choice and grammar reflect the witnesses’ experiences and thoughts, which are woven into their narratives. The structure of narratives and the sequencing of events are essential to determine the veracity of witness narratives (Pitcornell, 2013). For a narrative to be credible, witnesses must create a plausible sequence of events that correspond to the original events and communicate causality over time (Labov & Waletsky, 1967)[7].

Creating false narratives, such as witness statements, requires careful attention. While liars have time to plan, they must create narratives that alter events’ sequential and temporal order to create the illusion of truth. Deceptive language can vary and may appear vague, reticent, verbose, negative, subjective, or a combination thereof (Pitcornell, 2013). Furthermore, liars tend to embed deceptive elements into their otherwise truthful narratives, making deception detection more difficult (Loconte & Kleinberg, 2025)[8].

Prevaricators are more motivated to deceive when the consequences are severe, such as guilt, reputational damage, or fear of imprisonment. High stakes lead to clearer behavioral cues of deception. However, replicating authentic crime scenarios in a laboratory is unethical, hindering the study of deception under real-life conditions and leaving research findings challenging to interpret.

One theory of deception detection examines the cognitive processes involved in deception (DePaulo et al., 2003)[9]. Truthful individuals convey facts, whereas deceivers must not only remember facts but also monitor their verbal and nonverbal behaviors, as well as those of the person they are deceiving, to ensure that suspicion is not aroused. Due to this increased cognitive load, deceivers may exhibit verbal, nonverbal, and paralinguistic indicators of deception (DePaulo et al., 2003). Another theory focuses on the physiological changes often occur during deception, such as increased skin conductance, elevated blood pressure, and altered respiratory patterns (DePaulo et al., 2003). Given that the thresholds for these physiological changes vary among individuals, developing a more stable platform for examining deception may yield more definitive deception indicia.

Focusing on grammar structures reduces the dependence on cognitive processing and physiological cues that detect deception. The advantage of studying grammar structures is that deceptive and truthful people use the same grammar rules to construct sentences in both stressful and non-stressful environments. The only difference between truthful and deceptive narratives is the omission or obfuscation of the truth.

Words are the building blocks for sentences, and grammar rules serve as blueprints for sentence construction. People construct sentences according to a predetermined set of grammar rules, which are relatively stable in native English speakers and only vary slightly despite a wide variation in intellect, language competence, and vocabulary strength (Chomsky, 1972; Labov, 1992). The stability of grammar rules within the English language provides a more stable platform to study the similarities and differences of grammar structures in truthful and deceptive conditions.

Generative-Transformational Grammar

Generative-transformational language theory posits that humans are born with universal grammar principles, and language develops innately based on environmental factors rather than learned behavior (Chomsky, 2002[10], 1965)[11]. Transformational-generative grammar involves a set of rules, known as transformations, that relate to the components of a sentence. Chomsky refers to the underlying syntactic structure as deep structure, which is transformed into complex sentences at the surface structure level by adding words and punctuation. While all languages share a deep structure of verbs, nouns, and objects, their surface structures vary due to different transformations.

The transformational-generative grammar theory proposes that people can create valid sentences in their languages without formal training, regardless of cultural background (Chomsky, 2002). The grammar structure is consistent among native English speakers, with only minor variations despite differences in intellect and competence (Chomsky, 1972). Labov (1992)[12] observed that approximately 75% of utterances are well-formed, and less than 2% are ungrammatical after applying editing rules.

Text Bridges

Text bridges are grammatical structures that circumvent withheld information. Text bridges include: adverbial conjunctives, transitional words, and subordinating words. These grammatical devices facilitate smooth transitions between ideas and sentences (Forlini et al., 1990)[13].

Adverbial Conjunctions

Adverbial conjunctions encompass eight categories, forming transitions from one idea to the next idea (Table 1). For example, Mary went to the store, and then she went home. The adverbial conjunctive then connects the first complete idea, Mary went to the store, with the second complete idea, she went home.

Table 1

A Comprehensive List of Adverbial Conjunctions and Their Function

Addition	again, also, then, besides, equally important, finally, first, further, furthermore, in addition, in the first place, last, moreover, next, second, still
Comparison	also, in the same way, likewise, similarly
Concession	granted, naturally, of course
Contrast	although, yet, at the same time, despite that, even so, even though, for all that, however, in contrast, in spite of, instead, nevertheless, though, notwithstanding, on the contrary, on the other hand, otherwise, regardless, still
Emphasis	certainly, indeed, in fact, of course
Example or Illustration	after all, as an illustration, even, for example, for instance, in conclusion, indeed, in fact, in other words, in short, of course, namely, specifically, that is, to illustrate, thus, truly
Summary	altogether, finally, in brief, in conclusion, in other words, in short, in simpler terms, in summary, on the whole, therefore, to put it differently, to summarize
Time Sequence	after a while, afterward, again, also, then, as long as, at last, at length, at that time, before, besides, earlier, eventually, finally, formerly, further, furthermore, in addition, in the first place, in the past, last, lately, meanwhile, moreover, next, now, presently, second, shortly, simultaneously, since, so far, soon, still, subsequently, thereafter, too, until, when

Transitional Words

Transitional words connect themes and ideas or establish relationships (Forlini et al., 1990). For example, «It rained on Saturday; therefore, the picnic was canceled.» The transitional word, therefore, connects the idea that it rained to the idea that the picnic was canceled. Transitional words group into four basic categories: time, contrast, result, addition, or example (Forlini et al., 1990) (Table 2).

Table 2

A Comprehensive List of Transitional Words and Their Functions

Time	after, afterward, before, during, earlier, final, first, later, since, meanwhile, then, until
Contrast	however, in contrast, indeed, instead, nevertheless, on the contrary, on the other hand, yet
Result	as a result, because, consequently, on account of, so, then, therefore, and thus
Addition or Example	also, besides, for example, furthermore, in addition, moreover

Transitional words indicating time include after, afterward, before, during, earlier, final, first, later, meanwhile, since then, and until (Forlini et al., 1990). For example, «The class went on break at 10:00 a.m. During the break, I drank a cup of coffee.» The transitional word during draws a relationship between the idea of going on a break and the idea of drinking a cup of coffee. Transitional words indicating contrast include however, in contrast, indeed, instead, nevertheless, on the contrary, on the other hand, and yet (Forlini et al., 1990). For example, «We went bowling instead of going to the movies.» The transitional word, instead, contrasts the act of going bowling and the act of going to the movies. Transitional words indicating result include as a result, because, consequently, on account of, so, then, therefore, and thus (Forlini et al., 1990). For example, «Tom had no money, so he borrowed ten dollars from his brother.» The transitional word, so, connects the idea that Tom had no money with the result of borrowing money from his brother. Transitional words indicating addition or example include also, and, besides, for example, furthermore, in addition, moreover, and too (Forlini et al., 1990). For example, «Sue read the novel War and Peace before she saw the movie.» The transitional word before connects the idea of reading the book «War and Peace» with the action of going to see the movie. Subordinating Words Subordinating words connect independent clauses and dependent clauses. Independent clauses stand alone; dependent clauses cannot stand alone. For example, «After I went to the library, I went home.» The dependent clause, «After I went to the library,» cannot stand on its own. However, the independent clause, «I went home,» can stand on its own. Subordinating words signal dependent clauses. Subordinating words include after, although, as if, as long as, because, before, even though, if, in order that, since, so, that, then, though, unless, until, when, whenever, where, wherever, where, whenever, and while.

Linguistic Markers of Veracity

Text Bridges

The proposed study takes a unique approach to detecting deception. Instead of measuring verbal and nonverbal cues triggered by physiological changes or cognitive overload, this study examines the grammar structures people use during deception. Deceptive people who lie by omission must use grammar structures that allow them to construct a series of sentences that circumvent or bridge the information that deceptive people desire to withhold.

Most individuals who are deceptive do not invent entire stories but rather present factual information up to the point where they wish to hide details, omit the concealed information, and then continue with truthful statements (Ekman, 1992)[14]. At the juncture where individuals desire to hide details, they use a grammar structure or «text bridge» to bypass the omitted. Text bridges are linguistic features that may provide a grammatical basis for distinguishing between truth and deception or intentionally withheld information.

Grammatically camouflaging withheld information involves ensuring continuity in sentence construction. Both deceptive and truthful individuals use the same grammatical rules in sentence construction. Identifying specific grammatical structures can help pinpoint parts of the narrative where information may be withheld.

Sentences in English can be categorized into four basic types: declarative, imperative, interrogative, and exclamatory. This study focused on declarative sentences. Declarative sentences can be further divided into four forms: simple, compound, complex, or compound-complex (Forlini et al., 1990). A simple declarative sentence contains a subject and a verb, such as «Birds fly.» A compound declarative sentence contains two or more independent clauses (Forlini et al., 1990), such as «John read a book and he wrote an essay.» A complex declarative sentence consists of one independent clause and one or more subordinate clauses (Forlini et al., 1990), such as «Tom answered the phone when his wife called.» A compound-complex sentence consists of two or more independent clauses and one or more subordinate clauses, such as «The car ran out of gas, and the driver walked five miles because no one could drive him to the filling station.»

Temporal-Spatial Lacunae

Simple declarative sentences limit temporal-spatial gaps. A bank robber can say, «I robbed the bank» or «I did not rob the bank.» Compound sentences consist of two or more independent clauses, limiting these gaps. However, the conjunction «and» can introduce some temporal-spatial gaps as it may imply a sequence of events. Complex and compound-complex sentences create significant temporal-spatial gaps. For instance, in «When I came home from work, I found my wife dead,» there is a gap between coming home and finding the wife dead. This could be due to withholding activities during that time or skipping over critical actions, such as a violent altercation.

Subordinating clauses connect unequal but related ideas to form complex sentences (Forlini et al., 1990). Subordinating clauses begin with words such as after, although, as if, as long as, because, before, even though, if, in order that, since, so, that, then, though, unless, until, when, whenever, where, wherever, and while. Because subordinating clauses create temporal-spatial lacunae, liars can secrete information. Again, truthful people can use subordinating clauses as behavioral contractions to consciously or unconsciously withhold information that they consider lesser included activities of the larger activity in which they are engaged or simply omit the information because they consider the information to be irrelevant.

Transitional words create temporal-spatial lacunae wherein liars can conceal information. As with subordinating words, truthful people can use transitional words as behavioral contractions to consciously or unconsciously withhold information that they consider lesser-included activities of the larger activity in which they are engaged or simply omit the information because they consider the information to be irrelevant.

Adverbial conjunctions connect two complete ideas (Forlini et al., 1990). Adverbial conjunctions include accordingly, again, also, besides, consequently, finally, furthermore, however, indeed, moreover, nevertheless, otherwise, then, therefore, and thus (Forlini et al., 1990). Adverbial conjunctions create temporal-spatial lacunae wherein liars can conceal information. As with subordinating and transitional words, truthful people can use adverbial conjunctions as behavioral contractions to consciously or unconsciously withhold information that they consider lesser-included activities of the larger activity in which they are engaged or simply omit the information because they consider the information irrelevant.

Text bridges may not always indicate deception but show where writers withhold information. These omissions can be logically explained, such as skipping mundane details in a sequence of events. For instance, «then» in «I got up, and then I took a shower, and then I ate breakfast» signifies omitted activities like turning on the water or getting milk. Liars may use similar tactics to hide something. Text bridges help identify omitted information in narratives that might be relevant or trivial. Text bridges used at critical junctures during interviews or written narratives signal to interviewers that the interviewee intentionally or unintentionally withheld information. Through follow-up interviews, the interviewer must decide if the withheld information is important to the inquiry. If the withheld information is of no value, then the investigator can ignore the text bridge.

Liars Use Fewer Words

Research indicates deceptive individuals tend to use fewer words and include less relevant information in their statements. They often provide shorter descriptions of fabricated events because shorter stories are easier to remember. Additionally, due to not having firsthand experience, they typically use fewer descriptive details about the event in question (Vrij, 2000). When asked to repeat their accounts, they find recalling fewer facts easier than remembering many invented details (Vrij, 2000; Ekman, 1992; DePaulo et al., 2003).

Liars redact their written narratives to make their stories believable. Successful deception relies on this editing process. How liars edit their written narratives can provide clues to uncover the deception. Schafer (2007) examined the predictive value of grammar structures to distinguish truthful narratives from deceptive narratives. Study participants watched a video of a person shoplifting an item from a convenience store. The participants were asked to pretend they were the person in the video and write truthful and deceptive narratives describing their activities in the store. The study examined three variables: total words, text bridge ratio, and spontaneous negation ratio. The study results showed that the combination of the three variables correctly identified truthful narratives 87 percent of the time and deceptive written narratives 67 percent of the time. In a cross-validation study, the three variables total words, text bridge ratio, and spontaneous negation ratio predicted truthful narratives 89 percent of the time and deceptive narratives 76 percent of the time.

Schafer, Ekici, Young, Maldonado, and Karlins (2023)[15] sought to identify the grammatical differences between truthful and deceptive verbal narratives using the same methodology as Schafer (2007). The study combined text bridges and spontaneous negations to create a text bridge/spontaneous negation ratio, which returned statistically significant in discriminating between truthful and deceptive verbal statements. The study also demonstrated that deceptive verbal statements contained the minimizing word «just» at a significantly higher level than truthful narratives. When the variables text bridges, spontaneous negations, and the word «just» were combined into one ratio, it was found to be effective and statistically significant in discriminating truthful verbal narratives from deceptive narratives.

DeCicco and Schafer (2015)[16] studied three linguistic variables, text bridge ratio, spontaneous negation ratio, and fewer words, to discriminate between truthful and deceptive narratives written in Spanish by native Spanish speakers using the same methodology as Schafer (2007). The study results showed that deceptive narratives in Spanish contained significantly fewer words, higher text bridge ratios, and higher spontaneous negation ratios than truthful narratives.

Schafer, Wen, Ekici, and Young (2020)[17] examined the predictive value of grammar structures to differentiate truthful written narratives from deceptive written narratives in the Chinese language. The four variables examined were total character count, text bridge ratio, spontaneous negation ratio, and character strikeouts. The study results showed that deceptive narratives written in Chinese by native Chinese speakers contained significantly fewer words, higher text bridge ratios, higher spontaneous negation ratios, and more character strikeouts than truthful narratives. It is important to note that despite being structurally different, the Chinese language study produced similar results as the English language study. In support, Hwang, Matsumoto, and Sandoval (2016) found that statement analysis is applicable across the English, Chinese, and Spanish languages regardless of gender.

Text Bridge Ratio

Differentiating truthful from deceptive written narratives requires more than just surface comparison. In experimental settings, a person’s truthful statement can be directly compared to their deceptive one. However, side-by-side comparisons are rare in real life because deceptive statements often mimic truthful ones. If liars use fewer words, text bridge ratios might predict the veracity of statements in the absence of direct comparisons. This ratio is calculated by dividing the number of text bridges by the total words in a narrative. Since deceptive statements generally use fewer words, the number of text bridges to total words ratio may indicate deception reliably. Text bridge ratios could help assess the truthfulness of written statements without needing a truthful narrative for comparison.

Spontaneous Negation Ratio

Negations are defined by words like no and not, including contractions (Adams & Jarvis, 2006)[18]. They appear in responses to open-ended questions rather than closed-ended ones. There is a distinction between negations and spontaneous negations. For instance, in response to «Did you rob the bank?» both truthful and deceptive people would say, «No, I did not rob the bank.» This is a negation, not a spontaneous one. For open-ended questions, people should describe their actions instead of what they didn’t do (Sapir, 1996; Rudacille, 1994). For example, when asked about the states he visited, a tourist said, «I visited California, Utah, and Texas, but I never visited New York.» The phrase «but I never visited New York» is a spontaneous negation, indicating significance. Common spontaneous negations include «I don’t mean to interrupt,» «I’m not trying to be obnoxious,» and «I don’t mean to rain on your parade.»

Spontaneous negations can function similarly to text bridges by bridging information gaps. When individuals use spontaneous negations, they often do not specify the actions they performed. In response to open-ended questions, individuals should describe the actions they undertook rather than those they did not (Sapir, 1996; Rudacille, 1994). Spontaneous negations in open-ended questions may offer additional clues to distinguish between truthful and deceptive narratives, especially when combined with text bridges. Spontaneous negations will be expressed as a spontaneous negation ratio by dividing the total number of spontaneous negations in a narrative by the total number of words in the narrative.

Word Qualifiers

Most people are reluctant to lie outright, so they use Word Qualifiers to reduce certainty, weaken personal commitment, and make sentences appear less assertive. Deceptive statements appear truthful within the parameters established by Word Qualifiers. Some common Word Qualifiers are: probably, think, believe, this, that, kind of, like, maybe, perhaps, presumably, roughly, about, sort of, generally, and mostly. Word Qualifiers modify the meanings of nouns and verbs, allowing liars to equivocate.

Just

The word «just» is often used to minimize actions or their consequences. Someone might say, «I just went to the store,» when they actually engaged in other, more significant activities. This minimization can be a tactic to downplay their involvement or hide the full truth. This act of minimizing actions is a common trait that people utilize when trying to deceive.

The following hypotheses examine the efficacy of discerning truthfully written narratives from deceptively written narratives using grammatical structures:

Hypothesis 1: The five variables text bridge ratios, spontaneous negations, tense changes, word qualifiers, and use of the word «just» will discriminate truthful narratives from deceptive narratives.

Hypothesis 2: The five variables text bridge ratios, spontaneous negations, tense changes, word qualifiers, and use of the word «just» will discriminate truthful narratives from deceptive narratives at the same rate as the ID-3 computer software program.

Materials and methods. The study consisted of 133 participants. Participants were drawn from the student population studying law enforcement and justice administration at a mid-western university. Since university students must demonstrate minimum literacy skills upon enrollment, this population ensured that the study participants can read and write in English. The participants read and signed consent forms before participating in the exercise.

The participants were provided with the following verbal directions.

You read an article in the morning newspaper that an unidentified man was found murdered near the sculpture in front of the university library. The cause of death was blunt trauma. Write a false statement telling the police investigators that you are the person responsible for the murder of the unidentified man.

The hypothetical scenario presented was fictional; no murder occurred. The student volunteers wrote what amounts to deceptive statements or, in other words, wrote false confessions. The participants were given unlimited time to write their deceptive narratives.

A total of 133 statements were received (N = 133). The researchers transcribed each handwritten narrative verbatim using Microsoft dictation. The transcribed statements were then analyzed by the Incident Driven Deception Detection (ID-3) computer program. The ID-3 computer program uses a proprietary algorithm to detect the veracity of written statements. The ID-3 computer program comprises two central components: the computer processing program and a language-specific database. ID-3 uploads and analyzes text files such as witness and suspect statements, interview transcripts, trial transcripts, deposition transcripts, and other types of narratives for veracity. The database consists of words and phrases, referred to as variables that tend to indicate deception.

Measures

The independent variable condition has five levels — text bridges, spontaneous negations, tense change, word qualifiers, and use of the word «just.» The dependent variable has one variable – accuracy. The categories of deceptive and inconclusive were collapsed and treated as deceptive because an inconclusive score indicates that additional inquiry is necessary to determine the veracity of the narratives.

These independent variables are expressed as percentages.

The deceptively written narratives were scored as follows:

Text bridge percent -— Text bridge percent was calculated by dividing the total number of text bridges by the total number of words in the same narrative.

Spontaneous negation percent — Spontaneous negation percent was calculated by dividing the total number of spontaneous negations by the total number of words in the same narrative.

Tense change percent — Tense change percent was calculated by dividing the total number of present tense verbs by the total number of words in the same narrative.

Just percent -The Just percent was calculated by dividing the total number of times the word «just» was used by the total number of words in the same narrative.

Word qualifier percent – The word qualifier percent was calculated by dividing the total number of word qualifiers by the total number of words in the same narrative.

Results. The Logistic Regression Model was used to analyze the data. The dependent variable is the predicted accuracy of the transcribed narratives. The independent variables are the percent of the linguistic markers: text bridges, spontaneous negation, present tense, word qualifiers, and the word «just.» Refer to Table 3 for the results of the logistic regression.

Table 3

The results of the logistic regression

	Marginal Effect	Standard Error	z	P>|z|	[0.025	0.0975]
Text bridge percent	0.1742	0.021	8.280	0.000	0.133	0.215
Tense change percent	0.1357	0.026	5.263	0.000	0.085	0.186
Just percent	0.0255	0.057	0.447	0.655	-0.086	0.137
Spontaneous negation percent	0.2037	0.032	6.346	0.000	0.141	0.267
Word qualifiers percent	———	——-	——	——	——	——

Discussion. Three of the linguistic variables analyzed in this study, Text Bridges, Spontaneous Negations, and Present Tense, matched the ID-3 computer software program at a 90 percent precision rate. The three independent variables when combined, closely approximated the results of the ID-3 program.

This study offers a new paradigm for detecting deception in written narratives. This linguistic approach does not rely on induced detection apprehension, the fight or flight response, or cognitive overload, which can vary from one person to another. Deceptive people use the same grammar structures as truthful people. Liars must use grammar structures to circumvent intentionally withheld information. This study identified three grammar structures: text bridges, spontaneous negations, and tense change to identify where dissemblers intentionally withhold information. Identifying these variables guides investigators to the places in narratives where further investigation is required.

Relying on grammar structures to detect deception is a more stable platform because liars and truthful people must use a strict set of grammar rules to communicate the truth or a lie. Examining the grammatical differences between how people tell the truth and how they lie provides a new avenue to study deception, which may prove to be more reliable.

The unique feature of this linguistic approach is that the same approach to detect deception can be used in foreign languages, as demonstrated by Schafer (2007), DeCicco and Schafer (2015), Schafer, Wen, Ekici, and Young (2020), and Hwang, Matsumoto, and Sandoval (2016). This linguistic approach may provide a new method to detect deception. Nevertheless, further research is necessary to confirm this.

The results of this study serve to further assist investigators in detecting deceptive statements of witnesses, victims, suspects, and other persons of interest related to a law enforcement investigation. Investigators should be provided with additional training in detecting deception by focusing on the presence of linguistic markers, specifically Text Bridges, Spontaneous Negations, and Present Tense.

Conclusion. This study and its results were limited to college students as a sample population. Additional and varied population samples should be examined to determine if consistent results are obtained. This study analyzed deceptive statements; additional studies are warranted wherein truthful and deceptive narratives are analyzed. Narratives based on actual occurrences versus fictitious scenarios may garner different results.

 

References

1. Rabon D. Investigative Discourse Analysis. — Durham, NC: Carolina Academic Press, 1996.
2. Sapir A. (1987) Scientific Content Analysis, Phoenix: unpublished manuscript. Laboratory for Scientific Interrogation.
3. Kaster J. Interviewing Witnesses and Statement Analysis, unpublished manuscript. — Orleans, Ontario, Canada, 1999.
4. Rudacille W. C. (1994). Identifying Lies in Disguise, Dubuque, IO: Kendall/Hunt.
5. Steller M. & Koehnken G. Criteria based statement analysis in D. C. Raskin (ed.) Psychological Methods for Criminal Investigation and Evidence. — New York: Springer, 1989. — Р. 217-245.
6. Johnson M. K. & Raye, C. L. (1981). Reality monitoring. Psychological Review, 88, 67-85.
7. Labov W. & Waletsky J. (1967). Narrative analysis. In Helm, J. (ed.) Essays on the Verbal and Visual Arts. Seattle: University of Washington Press
8. Loconte R., & Kleinberg B. (2025). When lies are mostly truthful: automated verbal deception detection for embedded lies. arXiv preprint arXiv:2501.07217.
9. DePaulo B.M., Lindsay J.J., Malone B.E., Muhlenbruck L. Charlton K. & Cooper H.  Cues to deception // Psychology Bulletin. — 2003. — № 129(1). — Р. 74-118.
10. Chomsky N. Syntactic structures. — New York: Walter de Gruyter. 2002
11. Chomsk N.  Aspects of the theory of language. — Cambridge, MA: The MIT Press, 1965.
12. Labov W.  Language in the inner city. — Philadelphia: University of Pennsylvania Press, 1992.
13. Forlini G., Bauer M. B., Biener L., Capo L., Kenyon K. M., Shaw D. H., & Verner Z. (1990). Grammar and composition. Englewood Cliffs, NJ: Prentice Hall.
14. Ekman P.  Telling lies: Clues to deceit in the marketplace, politics, and marriage. — New York: W. W. Norton, 1992.
15. Schafer J., Ekici N., Young D., Maldonado K., & Karlins M.  Verbal indicators of deception // Journal of Forensic Sciences and Criminal Investigation. — 2023. — № 17. — Р. 1-9.
16. DeCicco A. J. & Schafer J. R. Grammatical differences between truthful and deceptive narratives // Applied Psychology in Criminal Justice. — 2015. — № 11. — Р. 76-92.
17. Schafer J., Wen M., Ekici N., & Young D. Detecting truthful and deceptive statements: An experiment on the Chinese language International. Scientific and Practference Proceedings. — Vladimir City, Russian Federation, 2020. Р. 95-107.
18. Adams S. H. & Jarvis J. P. () Indicators of veracity and deception: an analysis of written statements made to police // International Journal of Speech, Language and the Law. — 2006. — № 13 (1). — Р. 1-22.