How Lying Affects Linguistic Expression

29 May 2022

Have you ever been lied to?

The answer to that question is, of course, yes. You have likely even accepted lying as a fact of daily life. Lies come in all sizes and varieties, and for many people, social norms enforce simple lies in everyday conversation (How often are you actually doing well when you respond to your coworker’s “how are you?”). Given how pervasive lying is, one would expect it to be a relatively simple phenomenon, yet lying is actually a cognitively demanding task that arises from millions of years of brain evolutio


According to the social brain hypothesis, the increased brain size of humans evolved as a result of in-species social competition. As such, human evolution has emphasized a need for intelligence in order to outcompete other members of the species — particularly through an improved ability to lie and deceive. For example, in a communal society in which the spoils of a hunting and gathering are shared, it is beneficial for an individual to lie about how much food they find and save an additional portion for themselves. In this situation, a larger brain with a greater capacity for intelligence aids in survival and reproductive success. With such an evolutionary pressure to lie comes the pressure to detect lies in other

In order to outcompete others in a game of deception, it is critical that one can recognize when someone else is lying. Because of this, every person has an internal radar for lie detection, yet this system is subjective and easily fooled. As a result, various body-language techniques and lie-detection technologies, such as the polygraph, have been dedicated to reliably predicting whether an individual is lying or telling the truth. However, these methods can be subverted with proper training and will sometimes yield inaccurate results — hence why these tests are not admissible as evidence in the United States judicial system. Because of this, these tests are better used in fun situations or those with lower stakes, such as in this video from Vanity Fair featuring Kevin Hart taking a polygraph test.

A New Approach to Lie Detection

As the search for a reliable method of lie detection continues, one area of research has shown promise — using linguistics to analyze the structure of lies. According to Newman et al. (2003), “creating a false story about a personal topic takes work and results in a different pattern of language use.” When a person explains a past experience, personal belief, or any other personal subject, they are able to simply recall and share the memory. This process requires fairly little active attention, and allows the speaker to focus on other aspects of their explanation, such as providing reactions to events and using well-formed sentences.

However, if a person wishes to lie about one of these subjects, the cognitive demand is much higher. A liar is required to fabricate a believable story, separate this story from the truth, remember this story so as not to contradict it later, and provide the appropriate level of detail in their explanation to avoid contradicting the listener’s prior knowledge. Due to the cognitive load it causes, various patterns of speech tend to appear when a person lies. Thus, it is possible to derive a method for predicting truthfulness by analyzing these patterns compared to the manner in which a person speaks when they tell the truth.

In order to apply this method of lie detection, researchers have developed artificial intelligence text-analysis models to analyze differences between truthful and lying statements given by human subjects. The model used by Newman et al. (2003), known as Linguistic Inquiry and Word Count (LIWC), has been used to assess the truthfulness of novel statements using patterns of words in lying speech. In their experiment, the predictions of truthfulness made using LIWC yielded 61% accuracy, which is significantly more accurate than human assessment (Newman, 2003). In order to understand how LIWC accomplishes this task, one must understand the linguistic cues used by the model to assess lying. But first, what is even meant by lying?

Finding a definition of lying is a more complex task than it first seems. A first instinct is to describe lying as a mismatch between what is said and what actually occurred. However, this is complicated by the notion of implications. A person can say something explicitly true while simultaneously implying something false. For example, a person tells their friend “I totally stayed out all night,” when, in fact, they had stayed out at night. Explicitly, they told the truth to their friend — the words they used match the actual events. However, the speaker in this scenario utilized sarcastic emphasis, thereby implying that that they had not stayed out all night. Since the explicit statement does not match the implied meaning, the first definition cannot give a clear answer as to whether or not the speaker is lying.

Another definition of lying could be intentionally deceiving another person, where deceiving is defined as “deliberately leading someone into a false belief” (Meibauer, 2018). This definition, while closer to all-encompassing, still alienates some situations — particularly those in which the listener expects the speaker to lie. For example, if an notably untrustworthy person tells you “I saw your wallet in the garage and not the bedroom,” you would likely believe your wallet is in the bedroom. In this case, the speaker fails to lead you into a false-belief, and thus fails to lie to you under this definition.

In order to remedy these issues of definition, it is important to emphasize both the theory of mind and intentions of the liar, rather than the effects of the lie. A good working definition of lying for the context of linguistics comes from Williams (2002), where lying is defined as “an assertion, the content of which the speaker believes to be false, which is made with the intention to deceive the hearer with respect to that content.” This definition resolves the earlier counter-examples and allows for clear distinction between lies and non-lies. With this definition in mind, it is now possible to delve into the patterns that arise when a person lies.

Lying Pattern #1: Prosody

Given the cognitive demand of lying, a number of prosodic vocal cues associated with challenging thought processes appear when a person lies. According to Meibauer (2018), a person telling a lie naturally pauses more frequently and for longer durations while speaking, which gives them more time to formulate their lie. Additionally, the cognitive demand of lying causes a delay in responding to questions, more speech errors, and a slower than normal speech rate (Meibauer, 2018). Most interestingly, the pitch of a person’s speech is elevated while lying and continues to rise progressively at the ends of phrases (Meibauer, 2018). These patterns occur subconsciously as a person works to develop a lie and can be observed in real-time. In conjunction with body-language cues, prosody can give a good indication of whether a person is lying. However, a liar can train themselves to avoid presenting these cues, especially when telling a well-rehearsed lie. The following video by reiterates some of the speech cues present when a person lies:

Lying Pattern #2: Semantics

In order to observe more subtle linguistic characteristics associated with lying, it is useful to enlist the help of computers. LIWC, the program used by Newman et al. (2003), analyzes text and returns the proportion of the total words that fit under certain linguistic categories, such as third-person pronouns or negations. Comparing the proportions of word categories used in truthful statements with lying statements allowed the researchers to develop a model for predicting the truthfulness of independent statements using the weights of the word categories returned by LIWC after analyzing the statement (Newman, 2003).

By observing predictors used by this model, it is possible to discern which semantic word categories are more prevalent in the telling and writing of lies. According to Newman et. al (2003), the model suggests that lying is “characterized by the use of fewer first-person singular pronouns (e.g., I, me, my)[…] more negative emotion words (e.g., hate, anger, enemy), fewer exclusive words (e.g., but, except, without), and more motion verbs (e.g., walk, move, go).”

The reduced usage of first-person singular pronouns can be explained by a liar’s desire to distance themselves from their lie. By “investing less of themselves in their words” (Newman, 2003), liars are able to avoid feeling dishonest with themselves as a result of increased psychological distance. This lack of self-referential pronouns may also help reduce the feeling of ownership a liar has over their words.

In addition to the findings of the model developed using LIWC, a meta-analysis conducted by Zuckerman et al. (1981) found that the usage negative emotion words is a significant indicator of deception. It is believed that the usage of affectively negative words while lying is associated with the subconscious (or possibly even overt) guilt and anxiety felt by a person when lying. The experience these negative emotions is thought to be reflected in the semantic choices of a liar.

The use of exclusive words implies a separation between what is and what is not included in a story. It is thought that this added layer of complexity is limited by the cognitive load associated with lying (Newman, 2003). It is already mentally taxing to fabricate a story, making the fabrication of elements not included in the imagined events a more challenging cognitive task than when an individual is telling the truth. As a result, individuals typically utilize fewer exclusive words when telling a lie.

A similar explanation can explain the increased usage of motion verbs in lies. Motion verbs allow for simpler explanations than more complex verbs that involve intentionality and internal evaluations (Newman, 2003). By using motion verbs more frequently, liars free up cognitive resources to construct a believable lie.

In addition to the findings of Newman et al. (2003), Meibauer (2018) claims that “liars tend to use more negations and more generalizing terms such as always, never, nobody, or everybody than do serious speakers […] and have reduced lexical diversity (i.e., the number of different words in a statement divided by the total number of words used in that statement).” The use of negations and generalizing terms allows liars to avoid unnecessary specificity that may increase the cognitive demand of formulating their lie. Furthermore, the repeated usage of the same words helps a liar to maintain the continuity of their story, leading to a reduction in the lexical diversity of their statement.

The following video from TED-Ed utilizes examples to explain some of the semantic phenomena associated with lying:

t:s.ndn. test:

Write & Read to Earn with BULB

Learn More

Enjoy this blog? Subscribe to altamash999

1 Comment

No comments yet.
Most relevant comments are displayed, so some may have been filtered out.