In the United States, approximately 1.4 million traumatic brain injuries (TBIs) occur every year with 230,000 resulting in hospitalization and an estimated 80,000 resulting in long-term disability annually (Thurman, Coronado, & Selassie, 2007). The prevalence of TBI in the United States was estimated to be 1.1% of the U.S. civilian population at the beginning of 2005 or approximately 3.17 million people who are living with a long-term disability secondary to TBI (Zaloshnja, Miller, Langlois, & Selassie, 2008). America's armed forces engaged in conflicts in Iraq and Afghanistan have experienced TBIs from explosions of rocket-propelled grenades, improvised explosive devices, and land mines. The prevalence of TBI has been estimated to be more than 22% among injured service members (Warden, 2006). Diffuse brain damage caused by TBI often results in persistent physical, psychosocial, cognitive, and communicative problems.
Impaired discourse is the hallmark of post-TBI cognitive-communication disorder and, due to the central role discourse plays in everyday communication, impaired discourse abilities contribute to the participation restrictions that underlie the social isolation commonly experienced among individuals living with TBI. The extent of discourse impairments in individuals with TBI influences the diagnostic process, formulation of prognoses, and development of effective interventions for social reintegration.
Cognitive-communication disorders, including discourse impairments associated with TBI, may be related to a disruption of executive functions (Ylvisaker & Szekeres, 1989). More than any other cognitive process, executive function skills are linked to the success of community reintegration (Sohlberg & Mateer, 2001). Aspects of executive functions thought to be critical for effective communication are self-awareness and goal-setting, planning, self-directing and initiating, self-inhibiting, self-monitoring, self-evaluation, and flexible problem-solving (Ylvisaker & Szekeres, 1989). Impairments of executive functions are widely reported in the literature on TBI. For example, the Wisconsin Card Sorting Test (WCST), a commonly used index of executive function, has been correlated with measures of monologic discourse (monologues) among children and adults with TBI (Brookshire et al., 2000; Coelho, 2002).
Of the multiple and diverse sequela of TBI, memory difficulty has been reported as a leading subjective complaint from survivors and caregivers (Murray, Ramage, & Hopper, 2001). However, the relationship between memory deficits and cognitive-communication disorders remains unclear (Marsh & Knight, 1991). Importantly, working memory, which is a type of short-term memory that temporarily stores information while it is being processed, has been linked to discourse-production deficits following TBI. Modest correlations have been found between working memory and narrative discourse measures (Chapman, Gamino, Cook, Hanten, Li, & Levin, 2006; Youse & Coelho, 2005). Also, working memory has been associated with figurative language use and syntactic processing (Moran, Nippold, & Gillon, 2006; Turkstra & Holland, 1998), both of which also may affect discourse functioning.
Discourse genres are separated into two broad categories: monologic (monologues) and interactive. Monologues, which do not require interaction, encompass a number of genres including descriptions, narratives, and procedural and expository discourse (Cherney, 1998). Descriptive discourse entails the attribution of features and concepts of a stimulus. Narratives involve storytelling, either through story creation or story retelling. Procedural discourse includes explanations of a series of actions to perform a task. Expository discourse informs a listener of a topic through facts or interpretation and draws upon higher-level thinking skills, such as comparison and contrast, cause and effect, and generalization. In contrast, conversational discourse is interactive with participants alternating roles as speaker and listener to exchange ideas, thoughts, and feelings.
Because conversation is such a prevalent mode of human communication, it may be argued that it has greater ecologic validity—how well a measure relates to real-life situations—than monologic discourse and that, therefore, assessment of discourse among individuals with TBI should focus primarily on conversational discourse. However, monologic discourse also has ecologic validity, as everyday conversation often incorporates a narrative framework. Furthermore, storytelling often is embedded in social exchanges (Mar, 2004).
Monologic discourse may be more useful clinically than conversation. In a study of the relationship of discourse to functional outcomes after TBI, narrative and procedural discourse measures were better correlated than conversational discourse measures with social integration and quality of life (Galski et al., 2008). Four of the five significant predictors for social integration were monologic discourse measures related to efficiency, organization, and productivity. Additionally, quality of life was correlated only with monologic discourse measures. These findings suggest that conversational tasks may not be sufficiently difficult or cognitively demanding or may be too variable to use in the valuation of cognitive-communication abilities underlying successful community reintegration and life satisfaction.
Methodologically, narrative discourse presents the possibility of systematic and quantitative examinations not offered by conversational discourse. In addition, elicitation of discourse through the use of stories or procedures makes it possible to identify the elements of the targeted output that can serve as the standard for comparison. For example, a story retelling task requires all participants to provide an account of the same basic elements of the stimulus story. By contrast, conversational tasks do not provide an equivalent target against which an elicited interaction can be evaluated. Although both conversational and monologic discourse sample distinct aspects of communication, monologic discourse has methodological advantages over conversation and therefore may be better suited for predicting and monitoring recovery following TBI.
Monologic discourse affords multilevel analyses that provide a systematic way to gain insights into the nature of discourse deficits. Monologic discourse measures are organized at four levels: within-sentence, across-sentence, text-level, and story grammar. Within-sentence analyses, sometimes referred to as microlinguistic analyses, focus on lexical, semantic, and grammatical processes. Within-sentence analyses include sentential complexity, propositional analyses, verbal errors, productivity measures, and essential content units. Across-sentence measures, referred to as microstructural measures, involve measures of cohesion. Cohesion refers to inter-sentential organization that is established by meaning relations that connect one utterance to the next. Cohesive markers, such as anaphoric references (e.g., she, his), bridge meaning between sentences. Examination of cohesion may involve judgments of cohesive adequacy or identification of usage patterns.
Beyond the sentence level, text-level analyses are global measures that involve the whole discourse sample. Text-level analyses, sometimes referred to as macrolinguistic analyses, include gist summarization (extrapolation and refinement of the central meaning of a discourse sample) and coherence (thematic unity of the discourse sample measures). Local coherence reflects how well two sentences are thematically linked; global coherence reflects how well each sentence of the sample relates to the overall theme.
Story grammar analyses examine the over-arching discourse structure, known as the schema. Story grammar rules govern the schema and facilitate comprehension and production of a story, specifying logical connections between characters and events. The episode unit is central to measures of story grammar. An episode consists of an initiating event that provides the motivation for a character's goal-based behavior; an attempt on behalf of the character toward the goal; and a direct consequence, reflecting attainment or nonattainment of the goal.
Sensitivity of discourse measures to the subtle communicative deficits of TBI varies according to the analysis used. Findings associated with within- and across-sentence measures are mixed. For example, some studies examining cohesion have found significant differences between individuals with TBI and non-brain-injured individuals; others have found comparable performance among the groups. In contrast, text-level and story grammar analyses in the TBI literature have consistently demonstrated sensitivity, indicating that global discourse analyses may be more fruitful in understanding the relationship between discourse ability and cognition (Coelho, Ylvisaker, & Turkstra, 2005; Chapman et al., 2006; Moran & Gillon, 2010).
Neural Correlates of Discourse
Discourse is situated at the crossroads of language and cognition. Following TBI, linguistic abilities are generally spared (i.e., phonological and syntactic aspects of single words and sentences in the absence of context) and frank aphasias are relatively uncommon. As a result, the disruption of discourse ability is thought to have cognitive underpinnings. The systematic examination of discourse in relation to cognition may shed light on the processes involved in discourse production and on organization of the neural structures involved in the production and comprehension of discourse.
Imaging studies have implicated a variety of cortical areas that appear to be important for discourse production and comprehension, suggesting that discourse ability is widely distributed in the brain. Both medial and lateral prefrontal cortices (PFC), temporoparietal and anterior temporal regions, and posterior cingulate have shown activation during production of stories (Mar, 2004). The medial PFC is identified as a key site for regulation of motivation and is associated with discourse processes of sequencing and selection. Additionally, it has been implicated in theory-of-mind tasks, pointing to a potential role in inferencing. The lateral PFC has multiple efferent and afferent connections with all other lobes of the brain—temporal, parietal, and occipital (see Figure 1 [PDF]). Given such extensive circuitry, it is not surprising that the lateral PFC, particularly the dorsolateral region, has been shown to mediate a number of important cognitive processes, such as executive functions, working memory, information retrieval, and set shifting (Frattali & Grafman, 2005). Temporal sequencing of stories, which is thought to reflect working memory processes, also appears to be a function of lateral PFC. Some evidence suggests that lateral PFC mediates retrieval of episodic information.
Temporoparietal and anterior temporal regions appear to be responsible for mental inferencing, such as theory of mind. Although some areas of the temporal lobes are responsible for basic semantic processes, these areas are not the same temporal areas implicated in discourse. However, some researchers have raised the possibility that the temporoparietal and anterior temporal regions may be involved in more complex sentence processing (Mar, 2004).
The posterior cingulate is associated with multiple functions, such as updating schemas with new information, imagery, and retrieval of episodic information. Emotional stimuli also appear to engage the posterior cingulate, suggesting that the posterior cingulate may regulate the influence of emotions on memory-related processes. The diverse processes housed within the posterior cingulate point to its potential role as a site for mental simulation of discourse (Mar, 2004).
The extensive connections of the PFC to other areas of brain make it a likely candidate for control and regulation of goal-oriented functions, such as discourse ability. Several models of PFC functioning converge on the notion that the PFC is critical for automatic and controlled processing. For example, Grafman (1995) proposed that the PFC is responsible for plan-specific knowledge, represented as memory units called structured event complexes (SECs). SECs specify goal-oriented event information in sequence. They are dynamic representations such that experience modifies their information. For example, hosting a dinner party with the goal of having the entire meal prepared by a certain time activates the appropriate SEC for that task. Sequences for this particular SEC might include sending out invitations, buying groceries to meet the needs of the selected recipes, setting the table, and sequencing the cooking so that all the food is ready on time. The organizational framework of SECs is analogous to that of story grammar and scripts, suggesting that SECs may form the basis for knowledge representation in discourse (Frattali & Grafman, 2005).
Cognitive-communication impairments are characteristic of TBI and difficult to quantify because typical language batteries tend to be geared toward individuals with aphasia. These batteries generally assess linguistic form and content, abilities that are often spared following TBI. Despite the facility of their spoken language, individuals with TBI may have more difficulty in conversational exchanges than individuals without brain injury; they may be tangential, disorganized, and poorer at conversational turn-taking (Coelho, Liles, & Duffy, 1991; McDonald, 1993). Such individuals are good talkers but poor communicators, and often rely on their conversational partners to assume the burden of organizing or providing structure for the successful exchange of information.
Individuals with TBI also struggle with pragmatics and in particular "reading" the social context, which in large measure determines the effectiveness of a communicative exchange (Body, Perkins, & McDonald, 1999). Successful communication depends on conveying appropriate content in a way that encourages the communication partner to sustain an interaction. The social isolation of individuals following TBI reflects their difficulty with discourse engagement. Social communication has been shown to correlate significantly with life satisfaction, social integration, social productivity, and occupational participation (Dahlberg et al., 2006; Galski, Tompkins, & Johnston, 1998).
Standardized language assessment batteries may not be the best approach to assess social communication abilities as they cannot factor the context of communication into the results and thus tend to provide an inaccurate index of functional communication abilities. In a review of 84 standardized assessments recommended by speech-language pathologists in response to a survey, only 31 mentioned TBI in the test manual (Turkstra, Coelho, & Ylvisaker, 2005). Of those, seven were deemed reliable and valid according to measures established by the Agency for Health Care Research and Quality. The authors of the review remarked on the "striking absence of a test developed for the evaluation of communication in individuals with cognitive-communication disorders" (p. 219).
Discourse is complex use of language that depends on both linguistic and nonlinguistic cognitive abilities. Individuals with TBI are challenged to produce coherent and cohesive discourse, but no standardized or norm-referenced assessment can delineate the degree of difficulty they encounter. Non-standardized assessments serve a variety of functions, including determining competencies in domains for which there are no standardized tests as well as describing performance in the context of real-world settings and activities (Coelho, Ylvisaker, & Turkstra, 2005). Discourse analysis is non-standardized assessment. Discourse analyses are time-consuming to perform but yield information regarding linguistic, cognitive, and social functioning that can be helpful in designing customized interventions for individuals with TBI.
Implications for Treatment
As a group, individuals with TBI, particularly in the chronic stage (two or more years post-injury), demonstrate relatively preserved linguistic abilities. Impairments noted in their narrative discourse production are associated with problems at a more macro-level—what Glosser (1993) described as cognitive procedures for integrating linguistic and nonlinguistic knowledge in order to maintain conceptual, semantic, and pragmatic organization of discourse. There are a variety of cognitively based explanations for these breakdowns including difficulties with executive functioning (including working memory), memory (i.e., the ability to integrate prior world knowledge within ongoing discourse), and attention.
Such explanations suggest that intervention for discourse deficits may be more effective when directed toward the improvement of cognitive abilities rather than discourse alone. More research aimed at identifying the key elements of a successful rehabilitation program for individuals with TBI is critically needed, especially to address their discourse deficits as they persist and pose a significant barrier to successful community reentry.
Cognitive-communication problems create serious challenges to an individual's potential for social, vocational, and academic success. Therefore, management of individuals with TBI should incorporate assessment and treatment of discourse impairments.