Discourse coherence and cognition after stroke: A dual task study
Introduction
Discourse production is a complex task that involves retrieving information from memory, deciding which elements to include or exclude, remembering what has already been said, planning upcoming utterances, accounting for what the listener may or may not know, all while maintaining topic over time. It is not surprising then that a variety of cognitive processes are hypothesized to be integral to maintaining the multilevel organization involved in telling a story (Glosser, Brownell, & Joanette, 1993). For example, Alexander (2006) suggests that discourse production is a complex, goal-directed task that requires memory, planning, and sustained attention. The cognitive elements required for linguistic and nonlinguistic knowledge integration in discourse can best be described as macrolinguistic functions which organize discourse beyond the sentence level (Kintsch & Van Dijk, 1978). Global coherence is a macrolinguistic measure of the higher-level conceptual maintenance of topic across the discourse as a whole, while local coherence is the continued maintenance of content from one utterance to the next (Glosser and Deser, 1991, Glosser et al., 1992). The current study examined the relationship between global and local coherence and cognition in a nonaphasic, post-stroke population with mobility impairments during an attention-demanding task: talking while walking.
Studies examining the effects of dual tasks on walking have become increasingly common in the gait literature, with most studies focusing on older adults because of the potential for falls in this population under distracting conditions (e.g., Bootsma-van der Wiel et al., 2003, Liu-Ambrose et al., 2009, Shumway-Cook et al., 2007, Woollacott and Shumway-Cook, 2002). In general, these studies have shown decreases in gait speed when a secondary cognitive task is performed. For example, Shumway-Cook et al. (2007) examined walking under dual task conditions in four age groups ranging from middle age to 85+. Although all groups decreased walking speed when performing a talking task, 20% of the 75–84 age group and over 50% of those over 85 stopped walking while talking (Shumway-Cook et al., 2007). Several researchers have investigated the utility of tasks which track whether individuals stop walking while talking as predictive measures for falls. The most common talking task employed in these studies is the verbal fluency task, which require participants to name as many items from a specified category as possible while walking a short distance (Bootsma-van der Wiel et al., 2003, Camicioli et al., 1997, Pettersson et al., 2007, Shumway-Cook et al., 2007, Williams et al., 2006). Verbal fluency tasks have traditionally been classified as executive function tasks (Lezak, Howiesson, & Loring, 2004), and thus, may be considered cognitive tasks rather than language tasks by some researchers. Other “talking” tasks in the gait literature include saying the alphabet or alternate letters of the alphabet (Verghese et al., 2007) or saying ‘yes’ to one color stimulus and ‘no’ to an alternate color (Bowen et al., 2001) both of which have attentional and/or executive function demands (Suchy, 2009). The majority of these studies focus on gait and balance changes and fail to report the reciprocal effects of gait on the talking task.
Only a few studies investigating the effects of walking while talking have used actual discourse generation as the dual task in either elderly populations or older adults with stroke (e.g., de Hoon et al., 2003, Hyndman and Ashburn, 2004); however, it is rare for researchers to actually analyze the discourse output. Notable exceptions to this trend come from Kemper and colleagues who focus on the effects of motor tasks on different elements of language use (Kemper et al., 2003, Kemper et al., 2006, Kemper et al., 2009). Kemper et al. (2003) elicited discourse from younger and older adults using rhetorical prompts while participants walked, tapped their fingers in a complex pattern, or ignored background noise. They compared the cost of performing these dual tasks on different aspects of language use (e.g., fluency, sentence complexity, content), as well as on rate and “time on task” (i.e., percentage of time spent walking or tapping while talking). They found that older adults decreased the rate of speech (number of words per minute) in dual task conditions. In contrast, younger adults produced shorter, less complex sentences in dual task conditions, so that their dual task discourse appeared more like that of the older adults in the single task. A more recent study has replicated these findings using a digital pursuit rotary tracking task in which participants produced language samples while using a touch pad or trackball to follow a target that moved along an elliptical track (Kemper et al., 2009).
As discussed, the support for dual task interference from walking and talking is robust in healthy older adults. After a stroke however, even tasks such as walking or talking with no simultaneous distracting task can lose their automaticity and become more attentionally demanding (McCulloch, 2007). Doing two things at once then may be particularly challenging for stroke survivors who appear to have made functional recovery. Kemper et al. (2006) reported that when stroke survivors spoke on various topics while concurrently performing simple motor or selective ignoring tasks, performance on both tasks declined. Although the stroke survivors’ speech was similar to that of the healthy older adults on standard clinical measures when tasks were performed one at a time, the content, grammatical complexity and fluency of the post-stroke group was impaired during concurrent walking or finger tapping (Kemper et al., 2006).
Recently, Plummer-D’Amato et al. (2008) examined the interaction of gait variables with two cognitive tasks and a language task in nonaphasic, mobility-impaired stroke survivors. Tasks included discourse production using prompts from Kemper et al. (2003), a 1-back task, and a clock task requiring participants to report whether, for a given time, both clock hands would be in a specified half of the clock face. Participants walked for about 3 min around an oval track with and without the concurrent tasks. Participants were significantly slower during all dual task conditions than during single task walking. Furthermore, the discourse task produced larger dual task effects on gait speed, stride length, and cadence (steps/min) than either of the other two tasks. Additionally, the effects were bidirectional; during the dual task, stroke survivors’ narratives had more pauses and fewer words and utterances per narrative, but did not show the effects on grammaticality and complexity reported in Kemper et al. (2006).
Prior dual task studies examining walking while talking in post-stroke groups have analyzed discourse at the “microstructure” level, measuring variables such as content, grammaticality, and fluency (Kemper et al., 2006, Plummer-D’Amato et al., 2008). To our knowledge, no study has yet performed a macrolinguistic analysis of discourse performance during dual task in a post-stroke population. As stated above, a macrolinguistic analysis includes global coherence, the maintenance of each utterance's meaning to the discourse topic, and local coherence, the maintenance of one utterance's meaning to the preceding utterance (Glosser & Deser, 1991). The current study presents a secondary analysis of the narratives from the Plummer-D’Amato et al. (2008) study using macrolinguistic measures that investigate global and local coherence performance during single and dual task conditions.
Global and local coherence analyses have been performed in a variety of populations with neurogenic communication disorders (e.g., Ash et al., 2006, Coelho and Flewellyn, 2003, Dijkstra et al., 2002, Glosser and Deser, 1991, Hough and Barrow, 2003, Laine et al., 1998, Rogalski and Edmonds, 2008) and in normal aging (Glosser & Deser, 1992). Several studies have reported a connection between coherence and cognition. Glosser and Deser (1991) hypothesized that coherence was a function of cognitive processing based on findings that individuals with Alzheimer's disease and closed head injury, who had disrupted cognition, had degraded coherence compared with normal older adults and patients with fluent aphasia. Similarly, Ash et al. (2006) found that frontotemporal dementia patients with social comportment and executive function disorder were impaired in global coherence, which they attributed to impairments in organization, planning, and executive function. Though these studies have reported a link between cognition and coherence, none has directly tested the relationship between cognitive abilities and either local or global coherence.
In healthy older adults, Glosser and Deser (1992) postulated that cognitive processes such as executive function, working memory, and long-term memory are critical components of global coherence. A related line of research investigated “off-topic verbosity” in healthy older adults, and reported that older adults were more likely to produce more speech that was irrelevant to the topic than younger adults (e.g., Arbuckle and Gold, 1993, Arbuckle et al., 2000, Arbuckle et al., 2004, Gold and Arbuckle, 1995). Gold and Arbuckle (1995) demonstrated that off-topic speech did not correlate with vocabulary, but did correlate with cognitive measures which they claimed examined inhibitory function (i.e., Wisconsin Card Sort, Verbal Fluency, Trail-Making Test). However, other research has shown that the presence and severity of off-topic verbosity in older adults may be dependent upon the elicitation method. James, Burke, Austin, and Hulme (1998) reported that older adults were more likely to produce off-topic discourse in autobiographical narratives than in picture descriptions; whereas, Trunk and Abrams (2009) demonstrated that older adults were comparable to younger adults when producing autobiographical narratives but were rated as more off-topic when producing procedural discourse. The discrepancy between these studies may be accounted for by differences in scoring methods. When Trunk and Abrams calculated the percentage of off-topic words in a story both procedural and autobiographical discourse received higher off-topic verbosity scores, but when listeners rated the discourse on holistic story dimensions, only the procedural discourse was perceived as having greater off-topic verbosity. Consequently, there are indications that in certain types of discourse, older adults may produce more off-topic speech and this off-topic speech may be related to measures of executive function. However, exactly how off-topic verbosity relates to measures of global and local coherence is unknown. Thus, the evidence suggesting that off-topic verbosity correlates with measures of cognitive function does not address the issue of whether global and local coherence are related to cognitive abilities.
Participants in the current study were receiving or had completed therapy for gait impairments due to their stroke. We expected that the attentional demands of walking in this gait-impaired group would negatively impact discourse coherence in the dual task condition (walking and talking) more than in the single task condition (sitting and talking), because the dual task paradigm results in competition for cognitive resources (see Kahneman, 1973, McNeil et al., 1991), and coherence is hypothesized to involve cognitive processes. Furthermore, we predicted that global coherence would be more severely impaired by the dual task than local coherence, because global coherence requires the more attentionally demanding maintenance of topic over an extended period of discourse, while local coherence requires only maintenance between adjacent utterances.
In addition, we examined the correlation between different types of coherence, executive function measures, working memory measures, and vocabulary. While there is considerable face validity to the assertion that discourse coherence taps cognitive resources, it remains an empirical question. To our knowledge, this will be the first direct exploration of correlational relationships between different cognitive functions and measures of global and local coherence. Based on the literature, we made several predictions. First, we predicted that global coherence would correlate highly with measures of executive function, specifically those measuring inhibition and/or attention. The prediction regarding inhibition effects is based on studies of off-topic speech in normal aging which have suggested that inhibition of extraneous connections between ideas and events in memory is likely necessary for maintaining coherence throughout discourse (Arbuckle and Gold, 1993, Gold and Arbuckle, 1995). Additionally, it is possible that sustained attention may play an important role in maintaining a topic over an extended period of discourse (Alexander, 2006). With respect to local coherence, we predicted that it would correlate more highly with tests of working memory rather than executive function. Working memory requires holding and manipulating information for a short period of time (Baddeley, 1986); similarly, local coherence requires holding a preceding utterance in memory while constructing a subsequent utterance. Finally, we did not anticipate that there would be correlations between coherence and vocabulary measures. Vocabulary tasks require the access of declarative knowledge, that is, information that has been learned; in contrast, discourse coherence requires the continuous computation and updating of links to previous utterances as constrained by the global topic. Thus, there is little overlap in task demands between vocabulary tasks and the construction and maintenance of discourse coherence. Consistent with this reasoning, researchers have not found significant correlations between vocabulary and discourse measures in studies of off-topic speech (Arbuckle and Gold, 1993, Gold and Arbuckle, 1995).
Section snippets
Participants
Participants were 13 patients (11 males, mean age 60.46 years, range: 33–86) recruited from a homogenous community-dwelling stroke population (12 ischemic). All participants were within 2 years of having a stroke with the mean time post-onset being 8.69 months (SD: 4.79, range: 2.5–17 months). All were able to walk 10 m without physical assistance and could follow a three-step command. Participants did not have any pre-existing neurological disorders, severe hearing or visual impairments,
Results
All tests were considered statistically significant at p < .05. Global and local coherence scores were calculated as the sum of the coherence scores of each type for each T-unit divided by the number of T-units. A 2X2 within-subjects repeated measures ANOVA was conducted to determine the effects of task (single or dual) on coherence (global or local). No significant main effect was found for task [F(1,12) = .25, p = .625, partial η2 = .02]. However, there was a significant main effect of type of
Discussion
The present study used macrolinguistic measures, global and local coherence, to examine the effect of walking while talking in a gait-impaired group after stroke. Given the cognitive demands of discourse production and the attentional demands of walking, we predicted that coherence would be more affected during the dual task, walking and talking, than the single task, sitting and talking. We also predicted that global coherence would be more strongly correlated with cognitive measures,
Conclusions
In summary, we examined the relationship between coherence and cognition during a dual task, walking and talking, in a post-stroke population. Participants’ global coherence was poorer than local coherence regardless of single or dual task, and global coherence correlated with tests of sustained attention and speed of processing while local coherence did not. Findings suggest that global coherence may be more difficult to maintain than local coherence and that global and local coherence are
References (60)
- et al.
Effects of memory impairment on discourse
Journal of Neurolinguistics
(2000) - et al.
Individual differences in working memory and reading
Journal of Verbal Learning & Verbal Behavior
(1980) - et al.
Quantitative assessment of the stops walking while talking test in the elderly
Archives of Physical Medicine and Rehabilitation
(2003) - et al.
Conversational coherence: Discourse analysis of older adults with and without dementia
Journal of Neurolinguistics
(2004) - et al.
Mini-mental state: A practical method for grading the cognitive state of patients for the clinician
Journal of Psychiatric Research
(1975) - et al.
Patterns of discourse production among neurological patients with fluent language disorders
Brain and Language
(1991) - et al.
Speed and memory in the WAIS-III Digit Symbol—Coding subtest across the adult lifespan
Archives of Clinical Neuropsychology
(2004) - et al.
Coherence and informativeness of discourse in two dementia types
Journal of Neurolinguistics
(1998) - et al.
Interactions between cognitive tasks and gait after stroke: A dual task study
Gait & Posture
(2008) - et al.
Normative Symbol Digit Modalities Test performance in a community-based sample
Archives of Clinical Neuropsychology
(2006)
Analysis of normal discourse patterns
Brain and Cognition
Production of narrative discourse in aphasia
Brain and Language
Production of narrative and procedural discourse in aphasia
Brain and Language
The effect of elicitation task on discourse coherence and cohesion in adolescents with brain injury
Journal of Communication Disorders
Walking while talking: Effect of task prioritization in the elderly
Archives of Physical Medicine and Rehabilitation
Attention and the control of posture and gait: A review of an emerging area of research
Gait & Posture
Impairments of procedures for implementing complex language are due to disruption of frontal attention processes
Journal of the International Neuropsychological Society: JINS
Aging, inhibition, and verbosity
Journals of Gerontology
Effect of off-target verbosity on communication efficiency in a referential communication task
Psychology and Aging
Off-target verbosity, everyday competence, and subjective well-being
Gerontology
Trying to tell a tale: Discourse impairments in progressive aphasia and frontotemporal dementia
Neurology
Working memory
Walking and talking as predictors of falls in the general population: The Leiden 85-Plus Study
Journal of the American Geriatrics Society
Dual-task effects of talking while walking on velocity and balance following a stroke
Age and Ageing
Talking while walking: The effect of a dual task in aging and Alzheimer's disease
Neurology
Longitudinal assessment of coherence in an adult with fluent aphasia: A follow-up study
Aphasiology
My recaller is on vacation: Discourse analysis of nursing-home residents with dementia
Discourse Processes
Discourse production patterns in neurologically impaired and aged populations
A comparison of changes in macrolinguistic and microlinguistic aspects of discourse production in normal aging
Journals of Gerontology
Structural organization of discourse production following right hemisphere damage
Journal of Clinical and Experimental Neuropsychology
Cited by (38)
Impaired coherence for semantic but not episodic autobiographical memory in semantic variant primary progressive aphasia
2020, CortexCitation Excerpt :The focus of this study is on the relationship of the macro-structures of language, specifically deficits in global coherence, to semantic memory impairment within naturalistic autobiographical narratives. Other studies have previously employed coherence rating scales which vary in length, but invariably measure utterances on a spectrum ranging from coherent to incoherent (Glosser & Deser, 1990; Arbuckle & Gold, 1993; Kurczek & Duff, 2011; Reese et al., 2011; Race, Keane, & Verfaellie, 2015; Van Leer & Turkstra, 1999; Rogalski, Altmann, Plummer-D’Amato, Behrman, & Marsiske, 2010). To our knowledge, none of these rating scales developed so far has considered a score for utterances devoid of information and propositional content, such as fillers, rhetorical questions or repeated information.
The role of the right hemisphere in the recovery of stroke-related aphasia: A systematic review
2017, Journal of NeurolinguisticsCitation Excerpt :There is only weak evidence that age, sex, handedness and education influence functional recovery (Connor, Obler, Tocco, Fitzpatrick, & Albert, 2001; González-Fernández et al., 2011; Inatomi et al., 2008; Lazar & Antoniello, 2008; Lazar, Speizer, Festa, Krakauer, & Marshall, 2008; Lecours et al., 1987; Pedersen et al., 2004; Pickersgill & Lincoln, 1983). Pre-existing cognitive problems have a negative impact on aphasia recovery (Vukovic, Vuksanovic, and Vukovic, 2008), primarily because they affect the learning ability of the patients during rehabilitation (Rogalski, Altmann, Plummer-D'Amato, Behrman, & Marsiske, 2010). In contrast, a good insight in the language disorders, support from relatives and friends and a high motivation have a positive effect on the outcome (Basso, 1992; Koenig-Bruhin, Kolonko, At, Annoni, & Hunziker, 2013).
Cohesive and coherent connected speech deficits in mild stroke
2017, Brain and LanguageNeural substrates of spontaneous narrative production in focal neurodegenerative disease
2015, NeuropsychologiaCitation Excerpt :Consistent with past studies on narrative processing, storytelling was associated with a number of neuropsychological functions such as processing speed, working memory, and mental flexibility (Coelho et al., 1995; Mar, 2004; Marini et al., 2014). These studies have primarily examined the relationship between neuropsychological functions and narrative comprehension or depersonalized narrative production under prescribed experimental conditions, whereas this study is among few studies to examine the role of neuropsychology in spontaneous personal story production (see also Rogalski et al., 2010). We found that processing speed and mental flexibility (which in itself involves a combination of processing speed, working memory and task-switching abilities; Sanchez-Cubillo et al., 2009) were associated with the greatest number of storytelling characteristics.
Montreal–Toulouse Language Assessment Battery: Evidence of criterion validity from patients with aphasia
2015, Journal of the Neurological SciencesCitation Excerpt :No differences were found between patients with aphasia and those with RHD in the oral and written narrative tasks (IU and scene elements in both tasks, and the number of words produced in the oral narrative), or in the oral comprehension subtest. These results may be explained by the importance of the RH for pragmatic and semantic processing [30], and for complex discourse [50]. The oral text comprehension task also failed to distinguish patients with aphasia from those with RHD.
- 1
Present address: Department of Physical Therapy, Northeastern University, Boston, MA 02115, United States.