Linguistic correlates of self in deceptive oral autobiographical narratives

Consciousness and Cognition 20 (2011) 547–555

Contents lists available at ScienceDirect

Consciousness and Cognition journal homepage: www.elsevier.com/locate/concog

Linguistic correlates of self in deceptive oral autobiographical narratives J.S. Bedwell a,⇑, S. Gallagher b, S.N. Whitten a, S.M. Fiore b a b

Department of Psychology, University of Central Florida, Orlando, FL, USA Department of Philosophy, University of Central Florida, Orlando, FL, USA

a r t i c l e

i n f o

Article history: Received 20 January 2010 Available online 27 October 2010 Keywords: Autobiographical Narrative Self Deceptive Deception Coh-Metrix Cohesion Linguistic Explicit action verbs Referential coherence

a b s t r a c t The current study collected orally-delivered autobiographical narratives from a sample of 44 undergraduate students. Participants were asked to produce both deceptive and nondeceptive versions of their narrative to two specific autobiographical question prompts while standing in front of a video camera. Narratives were then analyzed with Coh-Metrix software on 33 indices of linguistic cohesion. Following a Bonferroni correction for the large number of linguistic variables (p < .002), results indicated that the deceptive narratives contained more explicit action verbs, less linguistic complexity, and less referential coherence (sentences being cohesive with each other). The results support a theory that, in deceptive narratives, there is greater narrative distance between the self that narrates and the self that is narrated about. This suggests that narrative selves are constituted not as autonomous selves, but are subject to processes (e.g., psychological, linguistic, social) that are likely operating on a subconscious level. Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction There has been an increasing amount of interest in narrative and the concept of narrative self in recent philosophy of mind and cognitive science (Dennett, 1991, 1992; Hutto, 2008; Nelson, 2003; Ricoeur, 1992; Schechtman, 1996, 2011; Velleman, 2006; Zahavi, 2007). The notion of a narrative self has been distinguished from other conceptions of self, and especially from the ‘‘minimal self,” defined as a pre-reflective consciousness of oneself as an immediate subject of experience, or what has been called the self-as-subject (Gallagher, 2000; Zahavi, 2005). Within the context of autobiographical (or self-) narrative, the narrative self is composed of both the narrating self (the narrator) who in the act of narration, ‘‘here and now,” is telling the story and is being affected by it, and the narrated self, the object (the protagonist) of the narrative. The dynamics between the narrating self and the narrated self can be studied in terms of the identity or narrative distance that exists between the narrator and the narrated self. ‘‘Narrative distance,” a concept that goes back to Aristotle’s Poetics, is used in narrative theory to indicate how far removed the narrator is from the narrated events (Andringa, 1996; Lothe, 2000). There are different kinds of narrative distance. (1) Perspectival distance: there is less distance between the narrator and the narrated events if the narration is done in the first person versus third person perspective. (2) Evaluative distance: narrative distance may also be measured in terms of the extent and the valence of the narrator’s evaluation of the events. (3) Temporal distance can be characterized as the distance between the time when the narrator narrates and the time represented by the narrated events. In this regard, if my narrative of a series of events is based on episodic memory, then the limitations imposed by my memory may introduce important ⇑ Corresponding author. Address: Department of Psychology, 4000 Central Florida Blvd., University of Central Florida, Orlando, FL 32816-1390, USA. Fax: +1 407 823 5862. E-mail address: [email protected] (J.S. Bedwell). 1053-8100/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.concog.2010.10.001

548

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

discrepancies between what I narrate and what actually happened – a difference between narrated events and historical events. Finally, (4) hermeneutical distance: narrative distance not only reflects differences in perspective, valuation, and temporality, it reflects how various interpretive aspects of the narration can introduce limitations and biases into the recounting of events. More generally, hermeneutic theory (Gadamer, 1989; Ricoeur, 1981) suggests that temporal distance is unavoidable and that all narrative recounting is an interpretation, the veracity of which is measurable in degree, but is never onehundred percent, due to factors such as the narrator’s interest or purpose, the audiences and their expectations, etc. Similar notions of distance hold for autobiographical (or self-) narrative. Specifically, one can ask about the distance between the self who narrates and the self who is narrated. When the narrator says, for example, ‘‘I had a great time in college,” the ‘I’ points in two directions. It points back to the person who is telling the story, the narrator, and it signifies that the narrator means to say something about herself; but it also points to the person or character who the narrator was at some point in the past. The narrator implies an identity between herself and the person she is talking about, but it is certainly a legitimate question to ask whether there is some degree of difference or narrative distance involved. First, even in a first-person autobiographical narrative, the person the narrator is today is not necessarily identical, in the relevant sense of ipseity or selfhood, to the person she was in the past.1 She might say, ‘‘I did this in college, but I certainly wouldn’t engage in that activity today. I’ve changed quite a bit.” Second, it’s also quite possible that the narrator doesn’t have it right when she states ‘‘I had a great time in college.” Perhaps she did not have a great time in college (as a historical fact) even if she is under the impression that she did. To some extent this will depend on the veracity or selectivity of her memory (perhaps she can only remember the few good things that happened); but it also will depend on certain hermeneutical biases generated, for example, by the kind of things she is interested in reporting, or by the kind of effect that she would like her narrative to have, or the kind of audience she is addressing. For these reasons it is not uncommon, in narrative theory, or in cognitive science, to find claims that autobiographical narrative involves a social or narrative construction of the self, or that some percentage of any particular self-narrative is a fiction or confabulation (e.g., Dennett, 1991; Flanagan, 1996; Gazzaniga, 1998; Jopling, 2000). In the study that we report here, we set aside these kinds of considerations, which generally go to the issue of the truth value of autobiographical narratives. We don’t mean to affirm (or deny) the truth value of the set of narratives that we’ll identify as non-deceptive. Our focus will be on narratives that are explicitly deceptive, the result of intentional misrepresentation or falsification (Ekman, 1985), and on contrasts between deceptive and non-deceptive narratives. At the same time, we think that the notion of narrative distance can be helpful in this respect. In addition this focus on explicitly deceptive narratives is different from the more controversial concept of ‘‘self-deception,” which includes deception that may or may not be consciously intended. Although there is ongoing debate about whether self-deception can be intentional (see, e.g., Bermúdez, 2000; Mele, 2000), and although the concept of narrative distance may be quite relevant to this debate, this is not something we need to address here. Again, the kind of deception involved in this study is clearly intentional misrepresentation in that the subject, following explicit instructions, sets out with the intent to create a deceptive narrative. In a study of non-autobiographical deceptive communication, Anolli, Balconi, and Ciceri (2003) point to evidence that in such narratives there is less self-reference, more frequent references to third parties and to objective factors, and greater use of impersonal vocabulary. They suggest that there is a greater impersonality or ‘‘depersonalization” (p. 689) involved in deceptive communications. This kind of depersonalization is consistent with an increase of narrative distance. In order to investigate the identity dynamics of the self in narrative, we examined syntactical structure in 176 short, oral autobiographical narratives, half of which resulted from instructions to make the narrative deceptive, and half of which resulted from instructions to make the narrative truthful (i.e., non-deceptive). We hypothesized that there would be an increase of narrative distance in deceptive autobiographical narratives compared to non-deceptive autobiographical narratives. That is, there would be a greater degree of impersonality or depersonalization in the case of deceptive than in non-deceptive narratives. We addressed this hypothesis with a comparative analysis of linguistic structure in the two kinds of narrative. 2. Materials and methods 2.1. Participants Fifty undergraduate students were recruited via academic credit incentives through the University of Central Florida Psychology Department’s online undergraduate research participation system. From this initial sample of 50 who chose to participate, four were excluded for self-report of psychotropic prescription use, as these medications have the potential to change cognitive functioning. From the remaining 46 participants, two were excluded as outliers for the total word length of their narratives (see Section 2.3 below). This resulted in a final sample of 44 participants who were included in the

1 Ricoeur’s (1992) distinction between idem-identity and ipse-identity is relevant here. Idem (sameness) refers to numerical identity, and in this sense one would say that the narrator is talking about himself rather than about someone else. Ipseity, however, is a concept that allows for a difference between myself as the one I am describing at an earlier (or later) time, and myself as narrator in the present. I may not be the same person I was when I was in college, X number of years ago. Narrative (self-) identity according to Ricoeur, is the product of the dialectic of idem and ipse.

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

549

analyses (71% female; mean age = 19.27, SD = 2.24, range = 18–32; 72.7% Caucasian, 15.9% African American, 9.1% Hispanic/ Latino, 2.3% Other/Mixed; 79.5% Freshman, 6.8% Sophomore, 9.1% Junior, 4.5% Senior). 2.2. Procedures Written informed consent was obtained from each participant prior to any research procedures. This study was approved by the University of Central Florida Institutional Review Board. Demographic information was then obtained for each participant, which included self-report of currently prescribed medications. A videotaped oral autobiographical language sample was obtained via two narrative prompts. These prompts were: (1) ‘‘Describe your immediate family during childhood, including the quality of your relationship with these individuals” and (2) ‘‘Describe your personality during high school, including examples of situations that occurred during this time period that exemplified your personality.” Prior to delivering the oral narrative for each prompt, the participant was given a written version of the prompt and was asked to prepare both a truthful and deceptive version for that prompt. Specifically, the researcher read the following script to the participant: For each question, you will be asked to tell two stories, one truthful story and one story that contains mostly lies. We will prompt you when to tell each type of story. In order to prepare for this task, please take this form and pencil and make some notes to prepare yourself for each of the stories you tell. For each story, keep in mind that you will be stopped if you exceed one minute during your actual videotaped story. You can speak for less than one minute, but please try to take up the majority of the minute with your story. Participants were also told that a blind rater would try to guess which stories were deceptive versus ‘‘truthful,” and to try and make the deceptive ones as believable as possible. However, in reality, there was not a blind rater that made these judgments, as this instruction was intended to motivate the participant to make realistic deceptive narratives. Participants were given four minutes to prepare the ‘‘truthful” and deceptive narratives for a given prompt, using written notes as needed. These notes were not allowed during the actual oral narrative delivery. For narrative delivery, the participant was asked to stand facing a camera, and was told to provide either the non-deceptive or deceptive version of the first prompt. This sequence of non-deceptive and deceptive versions was elicited in a counterbalanced order by participant (e.g., half were asked to tell the truthful story first). Participants were instructed to address the camera, with the experimenter seated behind them. The experimenter stopped any narratives that exceeded the one minute limit. The same preparation and delivery procedures were then used for the second of the two narrative prompts. The audio recordings of the narrative samples were then manually transcribed (including nonfluencies such as ‘‘um”) into text files. 2.3. Statistical analysis The total word length of each narrative was examined to identify participants with an unusually short narrative. We defined ‘‘unusually short” as the word length being less than two standard deviations below the mean for the entire group on that story. This procedure identified one participant who had unusually short narratives for both story types and another participant that had an unusually short narrative on one of the story types. We excluded these individuals from all further data analysis as it was possible that these individuals had difficulty with the task, which may have confounded the results from the group. After excluding these two participants, 44 remained for data analysis. The transcribed text files were imported into the software program Coh-Metrix 2.1 for analysis (http://cohmetrix.memphis.edu/cohmetrixpr/index.html). Coh-Metrix is a software program that automatically computes several linguistic features, or indices, of cohesion within a text (McNamara, Louwerse, Cai, & Graesser, 2005). The text analysis is conducted thorough a web interface which computes 56 linguistic features of the text (for a listing of these variables see: Graesser, McNamara, Louwerse, & Cai, 2004). Some examples of these linguistic features include word-level indices like familiarity, syntactic analysis, referential and semantic cohesion indices, and situation model indices (Graesser, Jeon, Yan, & Cai, 2007). The indices of text cohesion have been correlated with the subjective experience of coherence for the reader (Graesser et al., 2004). We first removed three linguistic variables: sentences per paragraph (we entered all oral discourse as a single paragraph) and two variables that examined syntactic structure in a manner that was very similar to the sentence syntax similarity variable that we chose for the analysis. The remaining 53 linguistic variables were then examined within each of the four conditions (two stories  deceptive vs. non-deceptive) for kurtosis and skewness. If any variable exceed a value of ±2.0 on both kurtosis and skewness for any of the four conditions, all four conditions of that variable were excluded from the remaining parametric analyses. This resulted in the exclusion of 19 of the linguistic variables, leaving 33 linguistic variables for use in the parametric analyses. A two (story type) by two (deceptive vs. non-deceptive) repeated-measures ANOVA was conducted on each of the 33 linguistic variables. In order to control for family-wise Type I error, we used a Bonferroni-corrected alpha of <.002 (.05/33). To further control for Type I error, we only considered the main effect of lie vs. truth if there was no interaction with story type (e.g., the deceptive effect was statistically similar across both stories). This also reduced the likelihood that the main effect of deceptive vs. non-deceptive conditions was not secondary to a particular story theme.

550

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

3. Results Seven linguistic variables showed a statistically significant main effect for deceptive vs. non-deceptive conditions and no statistically significant interaction with story type (see Tables 1 and 2). The variables and results are described as follows, using the example: ‘‘My sister was with me when the shuttle blew up. I pushed her away when I saw it.” 3.1. Flesch Kincaid Grade Level This is the traditional formula for calculating the grade level of textbooks. The higher the number, the harder it is to read the text. The grade levels range from 0 to 12. The formula is as follows: (.39  ASL) + (11.8  ASW) 15.59, where ASL = the average sentence length and ASW is the average number of syllables per word. In the example, this is 1.691. In the current study, the deceptive stories had a lower Flesch Kincaid Grade Level than the non-deceptive stories. 3.2. Number of sentences This is simply the number of sentences in each narrative. In the example, this is 2. In the current study, the deceptive stories contained more sentences than the non-deceptive stories. Notably, there was no suggestion of a difference in the number of words used between the deceptive and non-deceptive conditions, F(1, 43) = 0.31, p = .58, and no suggestion of a story type by deceptive/non-deceptive interaction on the number of words used, F(1, 43) = 0.74, p = .40. Therefore, in the current study, the deceptive studies contained more, but shorter, sentences than the non-deceptive stories – with no difference in the overall number of words used between the conditions. 3.3. Sentence syntax similarity The sentence syntax similarity index compares the syntactic tree structures of sentences. Coh-Metrix returns the proportion of nodes in the two tree structures that are intersecting nodes. This is an indication of how similar sentences are in structural complexity. In the example, this is 0.419. In the current study, the deceptive stories had greater sentence syntax similarity than the non-deceptive stories. 3.4. Intentional content (explicit action verbs) The incidence of intentional actions and events is estimated by counting the number of verbs that imply intentional, or goal-driven content. Intentional verbs represent explicit action: showed, put, made, pushed, learned, jumped. The Coh-Metrix terminology ‘intentional content’ is somewhat misleading because frequency of explicit action verbs does not fully capture everything that might fall under the concept of intentional content. In order to avoid the inference that these verbs necessarily reflect everything that might be counted as intentional (i.e., attributing intentional content to the person), we will refer to this variable as ‘‘explicit action verbs” for the purpose of this paper. This is precisely what is measured in the Coh-Metrix calculation. Coh-Metrix calculates incidence of explicit action verbs words per 1000 words. In the example, this is 55.556. In the current study, the deceptive stories contained a greater amount of explicit action verbs than the non-deceptive stories. 3.5. Log min in sentence for content words Returns the log word frequency value for the rarest word in each sentence and computes the mean across sentences. Scores range from 0 to 6. In the example, this is 2.546. In another example, the text: ‘‘My progeny was in an amiable mood and I found the day soporific. I was an astute learner and a formidable opponent. If however pedantic” has a Log min. freq of 0.385. However, the text: ‘‘She was in a good mood. I was great. If however mean.” has a Log min. freq of 2.42. Therefore, lower values relate to the general use of relatively rare words. In the current study, the non-deceptive stories had a lower value than the deceptive stories, indicating a greater use of relatively rare words in the truthful stories. Table 1 Linguistic variables showing a main effect of non-deception vs. deception with no interaction of story type.

*

Linguistic variable

F*

p

g2

Direction

Flesch Kincaid Grade Level Number of sentences Sentence syntax similarity Intentional content Log min in sentence for content words Stem overlap Anaphor reference

102.28 80.00 44.86 31.65 22.71 13.20 12.31

<.001 <.001 <.001 <.001 <.001 0.001 0.001

.71 .65 .51 .42 .35 .24 .22

ND > D D > ND D > ND D > ND D > ND ND > D ND > D

All F values represent the main effect of deception vs. non-deception from a two (story type) by two (deception vs. non-deception) repeated-measures ANOVA. All degrees of freedom are 1, 43. Under ‘‘direction,” D = deceptive condition and ND = non-deceptive condition.

551

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555 Table 2 Descriptive statistics for statistically significant linguistic variables. Variable by condition

Mean

Flesch Kincaid Grade Level Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

Standard deviation

Range

10.53 10.60 7.36 6.97

2.38 2.02 2.84 3.01

5.26–12.00 5.39–12.00 2.69–12.00 2.80–12.00

Number of sentences Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

4.23 4.80 8.36 8.89

2.81 2.19 3.46 3.83

1–11 1–10 1–18 1–17

Sentence syntax similarity Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

0.06 0.07 0.10 0.11

0.05 0.04 0.05 0.05

0–0.17 0–0.18 0–0.21 0–0.23

Intentional content Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

6.98 6.86 15.58 13.54

8.42 6.61 12.22 10.09

0–39.06 0–20.55 0–43.01 0–39.22

Log min. in sentence for content words Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

0.90 0.94 1.40 1.56

0.85 0.75 0.63 0.53

0–2.53 0–2.82 0–2.43 0–2.49

Stem overlap Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

0.42 0.55 0.33 0.36

0.37 0.30 0.23 0.22

0–1 0–1 0–1 0–0.90

Anaphor reference Non-deceptive – family story Non-deceptive – high school story Deceptive – family story Deceptive – high school story

0.63 0.78 0.52 0.60

0.39 0.29 0.22 0.28

0–1.00 0–1.00 0–1.00 0–1.00

3.6. Stem overlap Stem overlap is a form of referential cohesion, in which a noun in one sentence has the same or a similar morphological root as a word in another sentence. Coh-Metrix calculates the proportion of all sentence pairs in a paragraph that share one or more word stems. In the example, this is 0. Here is another example: ‘‘As the earth orbits the sun, it exerts a force. The orbit is not perfectly round.” ‘Orbits’ and ‘orbit’ have common stems, so there is stem overlap, even though one is a main verb and the other a noun. In the current study, the non-deceptive stories contained a greater degree of stem overlap than the deceptive stories. 3.7. Anaphor reference An anaphor is a word (usually a pronoun) that substitutes for a previously mentioned noun in order to avoid repetition. Coh-Metrix calculates the proportion of anaphor references that refer back to its constituent within five sentences. Anaphor reference is one aspect of referential cohesion. Referential cohesion occurs when a noun, pronoun, or noun-phrase refers to another constituent in the text. In the example, ‘her’ refers to ‘My sister’ and ‘it’ refers to the shuttle, resulting in two anaphoric references. In the current study, the deceptive stories contained fewer anaphor references than the non-deceptive stories. 4. Discussion Generally, the deceptive narratives showed: (1) less referential coherence (measured by Anaphor Reference and Stem Overlap) than the non-deceptive narratives. The deceptive narratives were also (2) less complex (lower Flesch Kincaid Grade Level, higher number of simple short sentences) with similar syntax (sentence syntax similarity) and less rare words (Log min. in sentence for content words). Importantly for purposes of this paper, the deceptive narratives also had (3) a significantly higher number of explicit action verbs.

552

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

Less referential coherence may reflect the fact that the participants are constructing the deceptive narratives on a more ad hoc basis than narratives based on memory of their own activities, which they may have described before. Despite the similarity in the total number of words, the deceptive narratives were also less linguistically complex. Notably, 71% of the variance between the stories in the Flesch Kincaid Grade Level was explained by whether the story was non-deceptive (more complex language) or deceptive (less complex language; see Table 1). We suggest that the higher number of explicit action verbs in the deceptive narratives reflects a greater narrative distance (a higher degree of impersonality) between the narrator and the narrated self. One way to think of the greater degree of impersonality (or depersonalization) in this context is to consider the self described in the narrative as more like another person (not me, not the true me). That is, if in fact I am lying about myself or about what I have done in the past, then in some sense I am describing someone who I am not. I’m describing someone other than myself. In that case I (the narrator) am giving a description of someone (myself – the narrated self) as another, and from the outside. In a first-person narrative the narrator does not know what goes on in the minds of the other characters except what is revealed through their actions, so in the descriptions of the other, actions become the main focus (Genette, 1983). Using this view, the prediction is that one’s narrations about people other than oneself will contain more explicit action verbs. That is, they will offer a description more in terms of actions and external events than in terms of the other’s motives or reasons for acting. Accordingly, the fact that our deceptive self-narratives contained more explicit action verbs suggests that the narrating self is distancing herself from the narrated self. The data also suggests that generating deceitful autobiographical narratives is more cognitively complex than generating non-deceitful narratives. Deceivers must use cognitive resources to maintain a coherent and consistent version of a false story, and to suppress ‘‘prepotent” responses, that is, to inhibit some aspect of the story that may correspond too much to the truth (Ganis, Morris, & Kosslyn, 2009; Johnson, Barnhardt, & Zhu, 2004). Moreover, there is an important difference in cognitive complexity between generating autobiographical narratives more on the basis of imagination (as in deceptive narratives) than on the basis of memory (as in non-deceptive narratives). In the case of memory, where I have a direct memory of my experience of the past event, I do not require an extra step of identifying who the subject is (Shoemaker, 1959). I know that it was I who experienced X. In the case of imagining myself do something, however, the structure is more like (1) I imagine someone does X; (2) I know that I did not do X; but (3) I nonetheless identify that someone as myself. Pretense enters into the deceit as an extra cognitive step. Some authors have argued that because lie telling is so routinely practiced, it may present only a slightly greater cognitive challenge than telling the truth (DePaulo et al., 2003). Notably, these same authors reported a non-significant small effect size for studies that have examined the construct of ‘‘cognitive complexity” as related to deceptive narratives (d = 0.07). However, the large majority of these previous studies did not examine deceptive autobiographical narratives, as were assessed in the current study. We argue that although individuals may have routine practice telling lies about isolated events and intentions, they do not tend to have much practice in telling deceptive autobiographical narratives about core aspects of their personality/self. Therefore, it is reasonable to conclude that deceptive autobiographical narratives would require significant cognitive resources, although the construct of cognitive complexity was not directly examined in the current study. The cognitive complexity involved in generating a false autobiographical narrative, however, doesn’t automatically translate into linguistic complexity. Indeed, as our study has found, the deceptive narrative may be expressed with a higher syntactical simplicity than the non-deceptive narrative. This, we propose, is a consequence of reduced cognitive resources available for language production during this type of deception. As a consequence of the increased cognitive load involved in deception, the narrator may be less able to engage his or her language production resources. Likewise, if the cognitive load is reduced (as in non-deceptive narratives), the narrator may have more language production resources available to enhance the story or embellish the details. There are conflicting findings in previous studies that have examined linguistic complexity in deceptive conversations. While one previous study was consistent with our finding of shorter sentences associated with deceptive communication (Zhou & Sung, 2008), more studies have found no differences in sentence length between deceptive and truthful conditions (Hancock, Curry, Goorha, & Woodworth, 2008; Zhou, Burgoon, Nunamaker, & Twitchell, 2004; Zhou & Zhang, 2006). One study found reduced punctuation (pausality) in deceptive communication, which was interpreted as reduced linguistic complexity (Zhou et al., 2004). Another study found reduced word length and noun-phrases related to deceptive communication in dyads, which was also interpreted as signs of reduced linguistic complexity (Zhou & Zhang, 2006). Similarly, an earlier study reported reduced use of unique words in deceptive narratives (Dulaney, 1982), which may be a sign of reduced linguistic complexity and is consistent with our finding of reduced unique word use in the deceptive narratives. However, one meta-analysis study found only a weak non-significant effect size for reduced use of ‘‘unique words” (d = 0.10) in deceptive narratives (DePaulo et al., 2003). We propose that these conflicting findings regarding linguistic complexity involved in deceptive narratives is related to our discussion about the conflicting findings on cognitive complexity in such narratives. Following our theory that reduced linguistic complexity is related to increased cognitive complexity, it may be that deceptive tasks that increase cognitive demands (e.g., autobiographical narratives) may be more likely to show reduced linguistic complexity. As the existing literature has used a wide variety of deceptive tasks which represent a wide range of relative cognitive challenge, this may explain inconsistencies in the report of linguistic complexity related to deception. In addition, while we acknowledge that individuals who are producing deceptive narratives are motivated to use language that is perceived to represent a believable and sincere utterance, it is not clear that individuals would believe that this requires more complex linguistic use or that they would have the capacity to self-monitor their language use during an oral deceptive narrative.

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

553

Bruner’s (1986) distinction between the landscape of action (LoA) and the landscape of consciousness (LoC) may help to explain both the idea that more explicit action words imply more narrative distance between narrating self and narrated self, and the idea that deceitful self-narratives are more cognitively complex. For Bruner, descriptions in the LoA describe what happens in terms of a series of actions or external events. In contrast the LoC describes ‘‘what those involved in the action know, think, or feel, or do not know, think, or feel” (Bruner, 1986: p. 14). Here is a simple example of the contrast between LoA and LoC in the context of self-narrative. LoA: I walked to the store to get some candy. LoC: I decided to walk to the store because I wanted to get some candy. Feldman, Bruner, Renderer, and Spitzer (1990) show, in a study of narrative comprehension in adults, that people easily understand the gist of a story cast solely in the less complex LoA, and it is clearly possible to represent the major events in a drama without always being able to decipher or explain, with full clarity or perhaps at all, the reasons why a protagonist will have acted. Generating LoA narratives require less cognitive work than LoC narratives, and may be preferred to balance out the higher cognitive demands of lying. Generally speaking, to describe my own actions or the actions of others (LoA) in simple terms is an easier task than to explain one’s own motives or reasons for acting (LoC). For example, as children gain narrative competency, they begin with narratives that are dominated by actions, and only gradually learn to add language that infers motives (Hutto, 2008; Nelson, 2009). And generally, to explain one’s own motives or reasons for acting may be an easier task than explaining the motives or reasons of another person. Clearly, however, describing another person’s motives, emotions, intentions, etc. in narrative form seems more complex than describing the actions of others. This is consistent with a previous report that found an increase in motion verbs (a subtype of action verbs) in deceptive narratives, which was interpreted in that study to reflect reduced cognitive complexity by similar reasoning (Newman, Pennebaker, Berry, & Richards, 2003). Furthermore, in line with the compensatory effect for the cognitive complexity of deception, noted above, the lesser degree of cognitive complexity for LoA narratives seems to be reflected in the lesser degree of syntactical complexity for the deceptive narratives (shorter and simpler sentences, lower reading level, syntactic similarity), consistent with what we find here. In regard to the question of narrative distance, where I treat myself more like another person, note that I have an external view on another’s behaviors – I see what the other person does, and I can tell you about his actions, but I may not be entirely sure about his motives. In describing the other person I can more easily tell you about his actions, or I can more easily imagine a set of actions to tell you about. If in the case of deceptive autobiographical narrative, I am describing myself as if I were another, and from the outside, then, we would expect the narrative to be framed more in the LoA than in LoC. This seems to be what happens in the deceptive narratives studied here, represented by the increased use of explicit action verbs. In summary, we propose that in deceptive narratives, there is greater narrative distance between the self that narrates and the self that is narrated about, than in non-deceptive narratives. Our results support this hypothesis, as reflected in the degree of impersonal structure – framed more in LoA than in LoC (more explicit action verbs) – and less syntactical complexity in the deceptive narratives. There is no claim made here that participants are aware that they are creating narrative distance between narrator and narrated self, even if they are clearly aware that they are creating deceptive narratives. If in fact they are not aware of this distance, it suggests that subjects are not fully in control of all aspects of their own narratives. Narratives are not entirely spun by us; rather, as Dennett (1991) suggests, narratives spin us, at least in part. Some philosophers (e.g., Ricoeur, 1981, 1992) would suggest that in the mélange of truth and fiction which make up self-narratives, which include both intentional deception and unintentional self-deception, narrative selves are constituted not as autonomous selves, but as subject to processes (psychological, linguistic, and social) that are often beyond their control. The results of this study are consistent with this view. This study, however, has a number of limitations that should be addressed in further research. First, one of the questions that remains unresolved is whether the deceptive narrator uses LoA (more explicit action verbs) and thereby creates more narrative distance because: (1) the narrator treats the narrated self as if it were someone other, or (2) it is a less cognitively complex strategy and it reduces the cognitive workload, or (3) some combination of (1) and (2).2 How we answer this question may determine some important features of the narrative distance involved. In the narratives under discussion here, we can rule out perspectival distance since all narratives were told in the first person. We can also rule out temporal distance since the narrative distance correlates to deceptive vs. non-deceptive rather than content, family (earlier) vs. high school (later). We might want to say, then, that if the narrator’s use of LofA is because the narrator treats the narrated self as if it were someone other, as in (1), the narrative distance involved may involve evaluative or hermeneutical components: evaluative, depending on

2 In a study that did not involve narrative, but simply responding, honestly or deceptively, to true or false statements about self vs. other, Ganis et al. (2009) found evidence that self-related deception and other-related deception activated different areas of the brain, as well as different regions in similar areas of the brain, e.g., different foci of activation within the insula. They suggest that the neural differences may be due to the fact that we are practiced in lying about ourselves, but less so in lying about others, and that there is a greater cognitive load (reflected in rate-limiting processes at the neural level) involved in lying about others than in lying about oneself. This does not address the question of whether being deceptive about oneself (vs. being honest about oneself) is more like treating oneself as another, but it does suggest that there may be some complex connection between greater cognitive load and treating oneself as another.

554

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

whether the narrated self presents a positive or negative picture; hermeneutical, in regard to what the narrator is trying to accomplish. However, if the narrator’s use of LoA is a less cognitively complex strategy to reduce the cognitive workload, this may involve something like the difference between use of memory (for non-deceptive) vs. use of imagination (for deceptive). In this case what seems to be narrative distance is simply an artifact of the cognitive requirements. Further research should be able to address this question. Furthermore, if it turns out that use of LoA is simply a strategy to reduce cognitive load, are there different strategies used in deceptive narratives that don’t involve LoA and that may reduce the appearance of narrative distance? There are several other limitations to this study. It considers only the syntactical aspects of a limited number of narratives generated in an experimental setting where subjects are asked to create deceptive narratives (see Sip, Roepstorff, McGregor, & Frith, 2008). It would be important to study ‘‘real life” deceptive narratives (e.g., perjured courtroom testimony), as well as narratives of greater length. ‘‘Real life” motivation for deception may have significant effects on the structure of deceptive self-narratives. Longer narratives would reveal their own overall narratological syntax which may not be reducible to the sentential syntax which was the focus of this study (see Petitot, 1995). Acknowledgments The authors would like to thank Micah Allen for his work with the data collection and data entry for this study and Scott Sutterby for assisting with the Institutional Review Board application. The authors would also like to thank Katie Ragsdale, David Richards, and Lauren Johnson for assistance with the linguistic analyses. The authors also thank Wendy Bedwell for her assistance with the task design. This study was funded by the National Science Foundation (BCS-0639037; PI: Shaun Gallagher). References Andringa, E. (1996). Effects of ‘narrative distance’ on readers’ emotional involvement and response. Poetics, 23, 431–452. Anolli, L., Balconi, M., & Ciceri, R. (2003). Linguistic styles in deceptive communication: Dubitative ambiguity and elliptic eluding in packaged lies. Social Behavior and Personality, 31(7), 687–710. Bermúdez, J. (2000). Self-deception, intentions, and contradictory beliefs. Analysis, 60(4), 309–319. Bruner, J. (1986). Actual minds, possible worlds. Cambridge, MA: Harvard University Press. Dulaney, E. F. (1982). Changes in language behavior as a function of veracity. Human Communication Research, 9(1), 75–82. Dennett, D. (1991). Consciousness explained. New York: Little, Brown, and Co.. Dennett, D. (1992). The self as a center of narrative gravity. In F. S. Kessel, P. M. Cole, & D. L. Johnson (Eds.), Self and consciousness: Multiple perspectives (pp. 103–115). Hillsdale, NJ: Lawrence Erlbaum Associates. DePaulo, B. M., Malone, B. E., Lindsay, J. J., Muhlenbruck, L., Charlton, K., & Cooper, H. (2003). Cues to deception. Psychological Bulletin, 129(1), 74–118. Ekman, P. (1985). Telling lies. New York: Norton & Company. Feldman, C. F., Bruner, J., Renderer, B., & Spitzer, S. (1990). Narrative comprehension. In B. K. Britton & A. D. Pellegrini (Eds.), Narrative thought and narrative language (pp. 1–78). Hillsdale, NJ: Lawrence Erlbaum Associates. Flanagan, O. (1996). Self-expressions: Mind, morals, and the meaning of life. New York: Oxford University Press. Gadamer, H. G. (1989). Truth and method (2nd ed.). London: Sheed and Ward. Gallagher, S. (2000). Philosophical conceptions of the self: Implications for cognitive science. Trends in Cognitive Sciences, 4(1), 14–21. Ganis, G., Morris, R. R., & Kosslyn, S. M. (2009). Neural processes underlying self- and other related lies: An individual difference approach using fMRI. Social Neuroscience, 4(6), 539–553. Gazzaniga, M. (1998). The Mind’s Past. Berkeley: University of California Press. Genette, G. (1983). Narrative discourse: An essay in method (J. E. Lewin, Trans.). Ithaca: Cornell University Press. Graesser, A. C., Jeon, M., Yan, Y., & Cai, C. (2007). Discourse cohesion in text and tutorial dialogue. Information Design Journal, 15, 199–213. Graesser, A. C., McNamara, D. S., Louwerse, M., & Cai, Z. (2004). Coh-Metrix: Analysis of text on cohesion and language. Behavioral Research Methods, Instruments, and Computers, 36, 193–202. Hancock, J. T., Curry, L. E., Goorha, S., & Woodworth, M. (2008). On lying and being lied to: A linguistic analysis of computer-mediated communication. Discourse Processes, 45, 1–23. Hutto, D. (2008). Folk-psychological narratives. Cambridge, MA: MIT Press. Johnson, R., Barnhardt, J., & Zhu, J. (2004). The contribution of executive processes to deceptive responding. Neuropsychologia, 42(7), 878–901. Jopling, D. A. (2000). Self-knowledge and the self. New York: Rutledge. Lothe, J. (2000). Narrative in fiction and film. Oxford: Oxford University Press. McNamara, D. S., Louwerse, M. M., Cai, Z., & Graesser, A. C. (2005). Coh-Metrix Version 1.4. Software. Retrieved 08.02.07. Mele, A. R. (2000). Self-deception and emotion. Consciousness and Emotion, 1, 115–137. Nelson, K. (2003). Narrative and the emergence of a consciousness of self. In G. D. Fireman, T. E. McVay, & O. J. Flanagan (Eds.), Narrative and consciousness: Literature, psychology, and the brain (pp. 17–36). New York: Oxford University Press. Nelson, K. (2009). Narrative practices and folk psychology: A perspective from developmental psychology. Journal of Consciousness Studies, 16(6–8), 69–93. Newman, M. L., Pennebaker, J. W., Berry, D. S., & Richards, J. M. (2003). Lying words: Predicting deception from linguistic styles. Personality and Social Psychology Bulletin, 29(5), 665–675. Petitot, J. (1995). How can physical symbols act upon semiotic structures and ‘‘visualized” meanings? Stanford Humanities Review, 4(1). . Ricoeur, P. (1981). Hermeneutics and the human sciences: Essays on language, action, and interpretation (J. B. Thompson, Trans.). Cambridge: Cambridge University Press. Ricoeur, P. (1992). Oneself as another K. Blamey, Trans.. Chicago: University of Chicago Press. Schechtman, M. (1996). The constitution of selves. Ithaca: Cornell University Press. Schechtman, M. (2011). The narrative self. In S. Gallagher (Ed.), Oxford handbook of the self. Oxford: Oxford University Press. Shoemaker, S. (1959). Personal identity and memory. Journal of Philosophy, 56(22), 868–882. Sip, K. E., Roepstorff, A., McGregor, W., & Frith, C. D. (2008). Detecting deception: The scope and limits. Trends in Cognitive Sciences, 12(2), 48–53. Velleman, J. D. (2006). The self as narrator. In Self to self: selected essays (pp. 203–223). Cambridge: Cambridge University Press. Zahavi, D. (2005). Subjectivity and selfhood: Investigating the first-person perspective. Cambridge: The MIT Press. Zahavi, D. (2007). Self and other: The limits of narrative understanding. In D. Hutto (Ed.), Narrative and understanding persons (pp. 179–202). Cambridge: Cambridge University Press.

J.S. Bedwell et al. / Consciousness and Cognition 20 (2011) 547–555

555

Zhou, L., Burgoon, J. K., Nunamaker, J. F., & Twitchell, D. (2004). Automating linguistics-based cues for detecting deception in text-based asynchronous computer-mediated communication. Group Decision and Negotiation, 13, 81–106. Zhou, L., & Sung, Y. (2008). Cues to deception in online chinese groups. In Proceedings of the 41st Hawaii international conference on system sciences. Zhou, L., & Zhang, D. (2006). A comparison of deception behavior in dyad and triadic group decision making in synchronous computer-mediated communication. Small Group Research, 37, 140–164.