BACKGROUND: RELATIONSHIP BETWEEN GAIT AND DISCOURSE
This discussion reports 3 pilot studies on the impact of different question types in conversation with persons who have cognitive impairments, both when seated and walking. These pilot studies respond to 2 major issues in long-term residential care for persons with cognitive impairment, particularly dementia of the Alzheimer type (DAT). One issue is exercise rehabilitation,1 since both gait and balance are affected in normal aging and even more so in DAT and other dementias. A second concern is providing social interaction for residents, primarily through conversation. Recently, 1 scholar2 has argued that conversation and exercise may delay deterioration in dementia. Our major focus will be on communication, particularly on conversational interaction.
Conversation forms the primary social interaction for most residents in long-term care, particularly for cognitively impaired residents in secure memory units of residential communities. Although there are, as yet, no national guidelines for developing communicative coaching for caregivers, our research is designed to contribute to such a project. Caregivers will benefit from learning that conversational acts such as direct questions (What did you/Who do you/When did you) call for access/retrieval, recall, and production of language, whereas other types of questions such as yes-no questions (Would you like this juice?) call for less-difficult processing of recognition, recall, and production.3 As Kempler4 summarizes, recognition tasks provide environmental support; recall tasks rely on internal mental processes, and aging affects recall far more than recognition.
Research suggests that increasing levels of physical activity produce favorable health outcomes among older adults and specifically those with osteoarthritis and other neurodegenerative chronic conditions.5–7 Interventions that combine physical activity with a cognitive component have been demonstrated to enhance function.8 Many researchers use dual-task protocols to examine the interaction of physical activity and cognition. Walking-while-talking paradigms9,10 are favored in many dual-task studies, particularly those focusing on gait and dementia. Researchers speculate that persons with dementia have problems with gait deriving from a decreased ability to shift attention or to prioritize cognitive resources for 2 concurrent tasks or some combination of these.11–13
According to Rogalski and colleagues,14 most studies of dual-task walking talking focus on gait and not on the impact that gait has on discourse. Discourse in DAT has been studied from different viewpoints. For example, literature concerning the Nun Study15 identified reduced idea density in the writing of persons with DAT long before the onset of symptoms. To examine idea density in conversational interaction in our 3 pilots, we used Computerized Propositional Idea Density Rater (CPIDR), a computer program measuring idea density in English16,17 by using these steps: (1) identify and count propositions (verbs, adjectives, adverbs, prepositions, and conjunctions) in a transcript of an utterance; (2) count the total words; and (3) divide the number of propositions by the number of words to obtain the idea density of an utterance.
Current research suggests ways to better measure interactions between walking velocity and other gait variables, as well as other tasks, which are needed for individuals with dementia.18 Counting19 or reciting lists in reverse order, the most prevalent modes for talk as a dual task, however, are not like everyday conversation. Conversation between care providers and residents during outdoor walks can be related to episodes of lucidity.20,21 Unfortunately, much of the conversation with residents with DAT consists of questions directed to the resident in an effort to elicit talk or of “talking over” the resident.22
Communication with cognitively impaired persons
Approaches to communication with individuals living with cognitive impairment present differing emphases and different kinds of discussions about how to ask questions. In a study of rehabilitating hip fracture in individuals with dementia,23 patients' problems with memory and insight were found to interfere with care. Recommended strategies included approaches drawn from relational communication.24 Relational communication is derived from studies of verbal/nonverbal communication in the 1980s by communication studies researchers. It was refined in 2001 into a scale evaluating the provider25 and is tied to both satisfaction and compliance.26,27 Question asking is seldom a focus. Caregivers are told to use reminders and simple instructions, reorienting patients to person, place, and time, using compensatory techniques such as calendars and other memory aids. Therapeutic communication, an approach drawn from techniques for enhancing conversations in therapy, is also emphasized in several areas of nursing. Examples of facilitators and barriers to therapeutic communication and to obtaining compliant behavior, often taken from a textbook first published in the 1960s,28 are abundant on the Internet, though not always credited to the original authors.
Two examples from current research show the range of approaches to language interaction and to questioning strategies. One study has tied therapeutic communication strategies to emergency care consultation,29 with an emphasis on what they call open-ended, affective questions. Their evidence was drawn from a qualitative study of videotaped consultations in which questions were predominantly 1 of 2 types. These types are “modal,” or content based, used by providers to elicit confirmation of biomedical detail, such as “this has happened before hasn't it,” and affective, such as “that was the last straw wasn't it,” to establish rapport and express sympathy. Open-ended questions (“What are you concerned about”) are recommended because they allow providers to elicit patient concerns.
A contrasting study defines therapeutic communication30 as an emotion-oriented approach emphasizing warmth, caring, and respect that can include behavioral management skills; it is recommended for training-assisted living staff, particularly unlicensed assistive personnel, and includes communication interventions such as asking 1 question at a time and using yes-no or 2-choice questions.
Attention to constructs such as the Communication Predicament Model,31 in which projections of incompetence on aging persons serve to bring about the very condition the older person hoped to avoid, has led to studies on the impact of patronizing speech and elderspeak,32 on the subtle part played by intergenerational speech,33 and on emphasizing retained abilities.34 Discourse approaches bring new insight into where and how the language of speakers with dementia can be studied to develop training. However, other than to recommend asking questions 1 at a time, little attention is paid to the type of questions that family members and caregivers apparently ask.
Question types in conversations with persons living with dementia
Because language processing in dementia is increasingly compromised35 as the disease progresses, therapists and caregivers are usually encouraged to use communicative strategies that include modifying or restricting question types.36 Question types used in conversations with persons having dementia (or similar cognitive impairments) have been studied according to whether they fall into either of the 3 categories: choice, yes-no, and open ended. Despite recommendations in the literature to use yes-no rather than open-ended questions, relying only on closed questions is not always useful because cognitively impaired residents are often able to respond to open-ended questions, as in conversations about feelings, when they cannot recall the specific information that is often called for by closed questions.37 Ripich and colleagues38 investigated caregiver use of open questions (WH questions, or questions beginning with who, what, when, where, why, which, whose, how), closed questions (yes-no questions: Would you like more coffee?), and choice questions (do you want this X or that X?), finding that without training, caregivers used more open questions. More recently, Small and Perry39 found that spouses used more yes-no questions, for fewer breakdowns, because yes-no questions reduced processing demands. Questions whose answers were drawn from semantic memory were more successful in eliciting responses than those requiring episodic memory, particularly of recent events: episodic memory is temporally graded, with more remote memory relatively spared in DAT40; episodic memory is related to a specific event and period of time; and semantic memory is time independent, reflects knowledge of the world, and is used over and over again. Semantic memory receives more rehearsal; it has a wider cortical network than autobiographical events, which, as episodic, are linked to the hippocampus and prefrontal lobes, which decline in normal aging and even more so in dementia. Petryk and Hopper41 have added an additional consideration: they analyzed whether responses were positive, neutral, or negative in terms of relationship to topic or content of question. They examined topic maintenance and conversation management and found a higher frequency of positive, on-topic utterances in response to open-ended semantic questions. Small and Perry39 exemplify how 3 basic question types (yes-no, choice, open) depend on different types of memory. The yes-no question “Do you want rice for dinner” calls on semantic memory, whereas the apparently similar yes-no question “Did we have rice for dinner last night” calls on episodic memory. Tag-questions (You want rice for dinner, don't you?) and so-prefaced questions (So, you have two sisters?) can also be answered with a yes or a no and can thus appear similar to choice or indirect questions.
In 3 pilot studies, we explored responses to 3 different closed questions plus an open question and omitted further study of choice questions because they were relatively rare outside of therapeutic discourse.42 The studies were as follows: (Pilot 1) “Scripted and unscripted conversation while sitting”: In this pilot, a single researcher administered 4 specific scripted questions during facilitative natural conversations, with 14 seated persons having high (experimental) and low (control) probability of dementia; (Pilot 2) “Unscripted, natural conversation”: Eight research assistants, 4 semitrained and 4 untrained, held unscripted, natural conversations with 4 seated persons having dementia. From this pilot, we present an analysis of the question types in the conversations by 2 research assistants with 1 person with DAT. (Pilot 3) “Scripted and unscripted conversation while walking”: This study examined the impact of a sudden, unexpected question, which had also been 1 of the 4 specific questions used in pilot study 1, while participants with DAT were walking on a GAITRite (CIR Systems Inc, Clifton, New Jersey) carpet and engaged in light conversation. A single researcher administered the sudden unexpected question to 13 persons (7 impaired and 6 unimpaired), while a second researcher studied the participants' gait. Full institutional review board review and informed consent procedures were implemented with participants in each study.
Pilot Study 1: Questions in scripted conversations while seated
Pilot 1 investigated differences in responses to questions that drew on semantic and on episodic-semantic memory. The study comprised 14 participants aged between 61 and 85 years, living in 2 long-term care communities. As is typical for assisted living communities, they had not been formally screened for or diagnosed with a particular form of dementia. Accordingly, we screened 18 participants for risk of cognitive impairment by using the 7-Minute Screen43,44 and recorded 7 with low probability and 7 with high probability of dementia. The remaining 4 were not available on the days when we were approved to visit and to do the recordings. All participants talked with the same researcher for 5 to 10 minutes, in their own residence, while seated. The conversations as a whole were not scripted: they began with small talk about the weather and went in any direction chosen by the resident. They always included a set of 4 scripted questions representing different question types: yes-no (Do you like roses), tag yes-no (You said you liked that shirt, didn't you), so-prefaced yes-no (So, you had two sisters), and an unexpected WH question unrelated to the current topic (What did you have for breakfast?). These 4 questions usually occurred in this order but were interspersed throughout the conversation.
The conversations were recorded on an Olympus DS 5000, transcribed in Transcriber software (available as free software under GNU General Public License, http://trans.sourceforge.net/en/presentation.php); pauses were analyzed in Praat (Praat, the Dutch word for “talk”) software, a gold standard for phonetic analysis by linguists and speech therapists.45 Pauses were annotated to the millisecond, using standard Praat settings. The 4 questions chosen for analysis comprised 3 different yes-no question types, which called on the recipient's semantic and semantic-episodic memory, and 1 WH question, which asked the recipient to draw on semantic-episodic memory:
When other questions occurred during the conversations, they were categorized and analyzed. We analyzed pause length46,47 to discriminate among question types, since pauses often signal processing and planning time.47
Data indicated that participants needed more time to respond to some question types than to others, as seen in Table 1. In general, WH questions (who/what/where/when/why/how) required longer response times. Tag-questions (You like this shirt, don't you?), yes-no questions (Do you like roses?), and so-prefaced questions (So, you had two sisters?) were usually answered faster than WH questions, such as “What did you have for breakfast?,” which was chosen to depart unexpectedly from the topic of discussion.
Table 1 also shows the differentiation in pauses by persons with low probability of dementia compared with persons with high probability of dementia. Overall, and for all question types except yes-no questions, participants with a high probability of dementia paused longer beforeresponding to questions than participants with low probability. In this data set, persons with low probability of dementia paused slightly longer before answering a yes-no question such as “Do you like roses?” In most cases, the persons with low probability went on to give a fuller response in terms of length of utterance and number of propositions in the reply than did the persons with high probability of dementia.
In addition, nearly all (12 of the 14) participants used a strategy of stalling for time to produce what they saw as an appropriate response to a WH question. Some simply paused. Others gave an immediate but inadequate response followed by a long pause before beginning the next, usually more adequate, part of the response, as in excerpt (a). Seven participants, all with high risk of dementia, also stalled after other questions, though usually for a shorter time, as seen in excerpt (b). Times in brackets are given in seconds.
Pilot Study 2: Questions in natural conversation while seated
The second pilot study comprised 8 graduate research assistants paired as conversation partners with 4 women, aged 61 to 85 years, with probable DAT, living in a secured memory care residence that was part of an assisted living community. As participants were living in assisted living (as opposed to a nursing home, which has different regulations), none had been formally screened for dementia, and staging was informally keyed to functional abilities requiring greater or lesser care by the care-giving staff. The memory care residence was divided into 3 units: least care, roughly equivalent to moderate cognitive decline (stage 4 on the Global Deterioration Scale);48 intermediate, roughly equivalent to moderately severe cognitive decline (stage 5); and greater care, roughly equivalent to moderately severe cognitive decline or to severe decline (stages 5 and 6, respectively). Two of the 4 women were in the least care unit and were fluent. One lived in the intermediate unit and was much less fluent. One lived in the greater care unit. She seldom produced utterances longer than 1 or 2 words, though she was capable of seeming engaged in conversation.
Pilot 2 focused on an inventory of the types of questions likely to surface in naturally occurring dementia care conversation and on the amount of information in the responses that each question type evoked. A report of the full project is in preparation. This initial analysis is a sample, consisting of the 10 conversations held with Maureen, a woman in the residence unit for those needing least care. Her student conversation partners (SCP) were Mina and Lorene, each of whom were asked to visit, converse, and record 5 conversations with Maureen over a 3-month period. All names have been aliased. Lorene had had some training in dementia care communication several months before the project; Mina had had none, though she had worked as a formal caregiver. The 2 SCP asked a total of 201 questions across the 10 conversations, 50% of which were yes-no questions. Neither SCP used tag questions. Question types were coded by 2 researchers, both linguists, who conducted discussion until consensus was achieved on the rare occasions they disagreed. Table 2 displays the types of questions asked by the SCPs in the sample.
Maureen answered all questions put to her and also asked them of her SCPs at appropriate stages of her conversations. Her responses were on topic, though not always particularly informative, and she could maintain topic for at least 2 and occasionally 3 turns. Excerpt (c) illustrates a typical response to a yes-no question, in this case, “Do you have any children?” In this utterance, a single dash represents a short pause:
(c) No I don't have any children I have uh uh some very wonderful uh nieces and nephews that have all done – one disciplined in nursing one uh then uh one one is air force retired colonel now and oh well just various you know things but they all of them have their own children so ah but um.
Maureen introduced new topics, such as the weather or inquiries about her SCP's life. In several instances, she showed awareness of the SCP's informational needs. She was also able to retain surface politeness and tact when asking 1 SCP to find another conversation partner by visiting someone else in the residence unit.
To assess how informative Maureen's responses to the questions were, we used the CPIDR program that determines the amount of information in an English text by measuring its propositional or idea density.16,17 The program was originally applied to schizophrenic and schizaphasic language production.49 Idea density, as previously described, is the number of expressed propositions divided by the number of words. Low idea density in written text can signal some kind of cognitive problem, such as dementia, as Snowdon found in the Nun Study and Kemper found elsewhere.15,16,50 Table 3 displays averages of Maureen's responses to each question type, sorted by the average number of ideas per response. The variability in the number of ideas and words in the responses can be seen in the high standard deviations (columns 4 and 6 of Table 3). However, the standard deviations for the average idea density (in the last column) are relatively lower, indicating that idea density is relatively stable across responses to question types, even though the number of ideas and the number of words used to express these ideas vary widely.
Like Small and Perry39 and Petryk and Hopper,41 we find the conventional distinction between open and closed questions less useful than that between episodic and episodic-semantic memory. We selected approximately the same amount of text, between 6500 and 7000 words from the conversations held by Maureen with each of her 2 SCPs, and looked specifically at only the 41 WH questions. Just less than half of the WH questions (19) were “closed,” rather than truly “open,” because in context, only a single answer could be expected. More importantly, 21 of the questions (including 12 of the 19 closed questions,) drew on both episodic and semantic memory rather than semantic memory exclusively. This may help in explaining why some WH questions are more difficult to answer than others.
Future studies will need to investigate changes in density as question types ask for increased information from episodic (combined with semantic) memory and should also investigate relationships between density and question type in longitudinal responses as well as cohort responses from persons who are less and more severely impaired.
Pilot 3: Questions in scripted conversation while walking
Pilot 3 examines 1 feature of a link between walking and talking in long-term care situations, with a long-term goal of contributing both to expanded social interaction and to falls prevention. Pilot 3 examined the impact of sudden questions on 13 residents, 7 cognitively impaired (D) and 6 unimpaired (N), while they were walking on a GAITRite Gold instrumented walkway obtained from CIR Systems Inc, Clifton, New Jersey. The walkway was 6.10-m long and was embedded with precalibrated sensors sampling at 80 Hz to capture all time-distance parameters of gait.51
The GAITRite was brought to the residence to minimize impact on an aging and cognitively impaired group of people as was earlier demonstrated by Camicioli.52 Talking while walking was obtained by using Olympus DS 330 digital recorders and a clip-on Sony microphone ECM 717. Participants walked 3 times across the carpet at their own speed,53,54 once without talking, once with facilitative talk consisting of ordinary observations about weather or clothing and/or a topic chosen or expanded by the participant, and once with facilitative talk into which was inserted an unexpected WH question unrelated to the current topic, “What did you have for breakfast?”
All utterances were transcribed by a trained graduate student and checked by the coinvestigators. Data for pausing were analyzed by a coinvestigator blinded to the data collection. Measurements of pauses were conducted by using transcriber and verified in Praat. Pauses between subject and researcher conversational turns were averaged to identify what could be considered the mean interspeaker pause length for each subject. Pauses encasing direct questions were annotated and averaged to identify what could be considered the mean pause lengths before and after the perturbation from the direct question. Speakers with dementia who used instant repetition of the question before responding, typically as a masking strategy, or “stall,” (“Hmm, what did I have for breakfast. Breakfast. Oh yes, I had oatmeal”) were checked by Maclagan, Davis, and a trained research assistant as an extra reliability check, with greater than 90% reliability. Since these speakers were all women, gender differences in responding to direct questions could warrant investigation along with other potential gender differences in early- and mid-stage Alzheimer disease.56
Typical pausing patterns for persons with dementia after the unexpected direct question, “What did you have for breakfast,” can be seen in excerpt (d)
Table 4 displays the averages by subjects of pauses between conversational turns and pauses after a direct question inserted into the conversation.
Average pause length between conversational turns was longer for people with dementia than for the controls. On average, both groups paused twice as long before unexpected WH questions, with the D(ementia) group's average pause before the WH question being relatively longer than the N(ormal control) group's average pause. In addition, pauses before the unexpected WH question were more than 2 SD longer than the average pauses between turns for 4 D participants but only 2 N participants (Table 4). Three of the controls, N1, N4, and N6, showed average pauses before WH questions that placed them closer to the dementia patients. This is unfortunately to be expected when controls are chosen from long-term care communities but cannot be administered specific screening.
The GAITRite and the Olympus recorder were synchronized manually rather than instrumentally. Nevertheless, analysis supported the identification of a subset of temporal-spatial variables that discriminated between both participants (unimpaired/impaired) and conditions (walking/walking while talking). Table 5 displays 6 time-distance parameters that can distinguish individuals with dementia from those of normal aging during walking, based on the current pilot data. These are standard variables used in gait analysis. The variables represent the output by using signal detection theory.53,54 Signal detection theory is a form of discriminant analysis that implements an iterative method for assessing the power of each time-distance variable to discriminate, individually and collectively as part of a model, the individuals with dementia from those who were normal in this study. All testing was based on α = 0.05; SAS 9.1 (SAS Institute Inc, Cary, North Carolina)56 was used. A full account of this study is the focus of another report, currently under review.
The speech analysis showed greater length of pauses before all question types for persons with dementia than without dementia, thus confirming the results of pilot 1. Both walking and talking, then, were disrupted by a sudden direct WH question, though other question types had appeared in the facilitative conversations and caused no such disruption.
The 3 pilot studies were designed to examine the impact of different question types on persons with and without dementia, in situations that promote conversation and social engagement, including sitting and walking. The long-range goal is to develop coaching interventions for caregivers and recommendations for clinicians. Talking en route to and from exercise, and even during an exercise program, allows formal and informal caregivers opportunity to increase social engagement and interaction, which is usually insuff icient for persons with dementia living in residential facilities. However, the way one talks to a person with dementia can have unlooked-for effects. The 3 pilot studies reported on here clearly indicate that some questions are more diff icult to answer than others. They also indicate that a simplistic division of questions into open (often equated with WH questions) and closed (similarly equated with yes-no questions) may not be helpful.
In pilot 1, findings suggest that pause length before responding to question type is correlated with probability or risk of dementia. Greater pause length was associated with higher probability or risk of dementia, and WH questions resulted in the longest preresponse pauses, presumably as people planned a response. In pilot 2, a case study of 1 speaker with DAT, we see that the majority of the questions asked of someone with dementia in natural conversation were yes-no questions and that the speaker's responses to yes-no questions carried an average amount of information in terms of both idea density and word frequency. Some WH questions received short answers with relatively little information, and others elicited longer responses with correspondingly more information. No tag questions were asked in unscripted conversation, even though they had the shortest preresponse pauses in pilot 1. Pilot 3 focused on walking and talking and found differentiation between persons with and without DAT, both in terms of gait and pause time after unexpected WH questions. As suggested by the greater pause lengths after direct questions for persons with dementia in pilots 1 and 3, some question types may not be beneficial in that they may affect attention and, if the person is walking, may impact gait.
A review of the literature on gait and cognition57 found that dual-task walking reduced gait velocity and increased gait variability in persons with DAT; it concluded that divided attention impaired the ability of a person with DAT to control variations in gait timing, which could increase chances of falling. A study of talking while walking with persons having DAT or mild cognitive impairment58 drew on earlier studies19,59 and supported their findings that the type of task while walking was important: first, the more difficult the postural control task, the greater the attentional demand; and second, in a talking-walking dual task, verbal fluency versus backward counting caused different gait-related changes.60 The 3 studies reported here are pilot in nature with obvious limitations in terms of sample size and generalizability.
The 3 pilot studies reported on here have implication for clinicians, because they clearly indicate that some questions are more difficult to answer than others.
- Clinicians need to be aware of the impact of question types. Questions that tap into detailed episodic memory about past events are likely to be more difficult for someone with dementia to answer (eg, “where did you live when you were at elementary school?”). Yes-no questions may be easier to answer (“do you like roses?”) because the listener has to only recognize the material, but they may elicit considerably less information from the person with DAT. Focusing exclusively on yes-no questions may lead to a 1-sided interaction with potentially unrelated questions eliciting very brief answers. Closed questions that tap semantic-episodic memory (“you really like roses, don't you?”) may actually facilitate longer responses. The person with DAT may respond with a simple yes/no, but they may also continue with additional comments. Open questions tapping semantic-episodic memory (“What do you enjoy doing in the afternoon?”) may also elicit longer responses.
- Clinicians and caregivers could benefit from training on “face” and its connection with identity.61 The typical adult response to a silent communication partner is to ask a question, but overemphasis on questions that ask a person with dementia to pull from a compromised episodic memory can cause embarrassment or even more negative emotions.
- Caregivers need coaching in the use of different question types and also in learning to comment on objects in the immediate environment and talking about areas the person is or was known to be interested in. This is more likely to elicit amicable conversation and social interaction than asking questions, even the most carefully chosen questions.
- In terms of walking, we note the identification of certain gait variables associated with individuals with dementia. We hypothesize that these variables could eventually be used by clinicians to track progression of the disease.
- Further investigation of the interaction of gait and everyday talk may allow clinicians and researchers to assess the effectiveness of interventions based on appropriate types of questions used to facilitate and train conversation between the caregivers and the patients.
- These pilot studies emphasize the value of multidisciplinary research on walking and talking in dementia. In particular, we advocate further research on question types and on walking and ordinary talking as a real-world set of tasks, with a focus on developing training from the findings for formal and informal caregivers.
In sum, dementia care communication requires more research involving real-world conditions, such as talking while walking, that can then be translated into dementia communication training techniques for clinicians and caregivers. Ideally, such techniques should foster person-centered care giving, as opposed to instrumental or task-oriented care; they should be translatable, presented in multiple media, using peer-reviewed or evidence-based techniques; they need to focus on outcomes emphasizing patient safety and quality of life; and they must be easily demonstrated for internalization by caregivers into daily routines, with subsequent brush-up62 or follow-up training.
1. Heyn P, Johnson K, Kramer A. Endurance and strength training outcomes on cognitively impaired and cognitively intact older adults: a meta-analysis. J Nutr Health Aging. 2008;12:401–409.
2. Arkin S. Language-enriched exercise plus socialization slows cognitive decline in Alzheimer's disease. Am J Alzheimers Dis Other Demen. 2007;22:62–77.
3. Kempler D, Almor A, Macdonald M, Andersen E. Working with limited memory: Sentence comprehension in Alzheimer's disease. In: Kemper S, Kliegl R, eds. Constraints on Language: Aging, Grammar and Memory. Boston, MA: Kluwer; 1999.
4. Kempler D. Neurocognitive Disorders in Aging. Thousand Oaks, CA: Sage; 2005.
5. Heyn P. The effect of a multisensory exercise program on engagement, behavior, and selected physiological indexes in persons with dementia. Am J Alzheimers Dis Other Demen. 2003;18:247–251.
6. Ettinger W, Burns R, Messier S, et al.. A randomized trial comparing aerobic exercise and resistance exercise with a health education program in older adults with knee osteoarthritis. JAMA. 1997;277:25–31.
7. Hughes S, Williams B, Molina L, Bayles C, et al.. Characteristics of physical activity programs for older adults: Results of a multisite survey. Gerontol. 2005;45:667–675.
8. Della Sala S, Cocchine G, Logie R, Allerhand M, MacPherson S. Dual task during encoding, maintenance and retrieval in Alzheimer's disease. J Alz Disease. 2010;19:503–515.
9. Sheridan P, Hausdorff J. The role of higher-level cognitive function and gait: Executive dysfunction contributes to fall risk in Alzheimer's disease. Dem and Geriatric Cog Disorders. 2007;24:125–137.
10. Verghese J, Kuslansky G, Holtzer R, et al.. Walking while talking: effect of task prioritization in the elderly. Arch Phys Med Rehal. 2007;88:50–53.
11. Woolacott M, Shumway-Cook A. Attention and the control of posture and gait: a review of an emerging area of research. Gait Post. 2002;16:1–14.
12. Verghese J, Buschke H, Viola L, et al.. Validity of divided attention tasks in predicting falls in older individuals: a preliminary study. J Am Geriatr Soc. 2002;50:1572–1576.
13. Sheridan P, Hausdorff J. The role of higher-level cognitive function in gait: executive dysfunction contributes to fall risk in Alzheimer's disease. Demen Geriatr Cog Disord. 2007;24:125–137.
14. Rogalski Y, Altmann L, Plummer-D'Amato P, Behrman A, Marskiske M. Discourse coherence and cognition after stroke: a dual task study. J Com Disord. 2010;43:212–224.
15. Snowdon D, Kemper S, Mortimer J, Greiner, Wekstein D, Markesbery W. Linguistic ability in early life and cognitive function and Alzheimer's disease in late life: findings from the Nun Study. JAMA. 1996;275:528–532.
16. Brown C, Snodgrass T, Kemper S, Herman R, Covington M. Automatic measurement of propositional idea density from part-of-speech tagging. Behav Rsch Meth. 2009;40:540–545.
17. Covington M, He C, Brown C, Naci L, Brown J. How complex is that sentence: a proposed revision of the Rosenberg and Abbeduto D-Level Scale. CASPR Research Report; 2006. University of Georgia Artificial Intelligence Center. http://www.ai.uga.edu/caspr
. Accessed May 5, 2011.
18. van Iersel M, Hoefsloot W, Munneke M, Bloem B, Olde Rikkert M. Systematic review of quantitative gait analysis in patients with dementia. Z Gerontol Geriat. 2004;37:27–32.
19. Beauchet O, Dubost V, Aminian K, Gonthier R, Kressig RW. Dual-task-related gait changes in the elderly: does the type of cognitive task matter? J Mot Behav. 2005; 37:259–264.
20. Normann H, Henriksen N, Norberg A, Asplund K. Lucidity in a woman with severe dementia related to conversation: a case study. J Clin Nurs. 2005;14:891–896.
21. Normann H, Asplund K, Karlsson S, Sandman P, Norberg A. People with severe dementia exhibit episodes of lucidity. J Clin Nurs. 2006;15:1413–1417.
22. Davis B, Pope C. Institutionalized ghosting: policy contexts and language use in erasing the person with Alzheimer's. Lang Policy. 2010;9:29–44.
23. McGilton K, Wells J, Davis A, et al.. Rehabilitating patients with dementia who have had a hip fracture II. Top Ger Rehab. 2007;23:174–182.
24. Gallagher T, Hartung P, Gregory S. Assessment of a measure of relational communication for doctor-patient interactions. Pat Ed Couns. 2001;45:211–218.
25. Burgoon J, Pfau M, Parrott R, Birk T, Coker R, Burgoon M. Relational communication, satisfaction, compliance-gaining strategies, and compliance in communication between physicians and patients. Commun Monographs. 1987;54:307–324.
26. Finch L. Patients' communication with nurses: Relational communication and preferred nurse behaviors. Intl J Human Caring. 2006;10:14–22.
27. Gallagher T, Hartung P, Gerzina H, Gregory S, Merolla D. Further analysis of a doctor-patient nonverbal communication instrument. Pat Educ & Couns; 57:262–271.
28. Hays J, Larson K. Interacting With Patients. New York, NY: Macmillan; 1963.
29. O'Gara P, Fairhurst W. Therapeutic communication part 2: strategies that can enhance the quality of the emergency care consultation. Accid Emerg Nurs. 2004;12:201–207.
30. Levy-Storms L. Therapeutic communication training in long-term care institutions: recommendations for future research. Pat Educ Couns. 2008;73:8–21.
31. Ryan E, Giles H, Bartolucci G, Henwood K. Psycholinguistic and social psychological components of communication by and with the elderly. Lang Comm. 2008;6:1–24.
32. Williams K, Herman R, Gajewski, Wilson K. Elderspeak communication: impact on dementia care. Am J Alzheimers Dis Oth Demen. 2009;24:11–20.
33. Williams A, Nussbaum J. Intergenerational Communication Across the Lifespan. Mahwah, NJ: Laurence Erlbaum; 2001.
34. Davis B. Alzheimer Talk, Text and Context: Enhancing Communication. New York, NY: Palgrave; 2005.
35. Small J, Kemper S, Lyons K. Sentence repetition and processing resources in Alzheimer's disease. Brain Lang. 2000;75:232–258.
36. Small J, Gutman G. Recommended and reported use of communication strategies in Alzheimer caregiving. Alz Dis Assoc Disord. 2002;16:270–278.
37. Tappen R, Wiliams-Burgess C, Edelstein J, Touhy T, Fishman S. Communicating with individuals with Alzheimer's disease. Arch Psychiatr Nurs. 1997;11:249–256.
38. Ripich D, Ziol E, Fritsch T, Durand E. Training Alzheimer's disease caregivers for successful communication. Clin Gerontol. 1999;21:37–56.
39. Small J, Perry J. “Do you remember?” How caregivers question their spouses who have Alzheimer's disease and the impact on communication. J Spch Lang Hear Rsch. 2005;48:125–136.
40. Small JA, Sandhu N. Episodic and semantic memory influences on picture naming in Alzheimer's disease. Brain Lang. 2008;104:1–9.
41. Petryk M, Hopper T. The effects of question type on conversational discourse in Alzheimer's disease. Persp Neurophys Neurogen Spch Lang Disord. 2009;19:126–134.
42. Crystal D, Fletcher P, Garman M. The Grammatical Analysis of Language Disability. London, England: Arnold; 1976, 1999.
43. Solomon P, Pendlebury W. Recognition of Alzheimer's disease: the 7 Minute Screen. Fam Med. 1998;30:265–271.
44. Solomon P, Hirschoff A, Kelly B, et al.. A 7 minute neurocognitive screening battery highly sensitive to Alzheimer's disease. Arch Neurol. 1998;55:349–355.
46. Han S, Kim M, Kim S, Kang H. Spontaneous speech rate in patients with mild cognitive impairment and early stage Alzheimer's disease. Poster P3-106, International Congress onAlzheimer's Disease. Alz and Dementia. 2010;6:S481.
47. Davis B, Maclagan M. Examining pauses in Alzheimer's discourse. Am J Alz Disease Oth Dementias. 2009;24:141–154.
48. Reisberg B, Ferris S, de Leon M, Crook T. The global deterioration scale for assessment of primary degenerative dementia. Am J Psychiatry. 1982;139:1136–1139.
49. Covington M, Riedel W, Brown C, et al.. Does ketamine mimic aspects of schizophrenic speech? J Psychopharmacol. 2007;21:338–346.
50. Kemper S, Sumner A. The structure of verbal abilities in young and older adults. Psychol Aging. 2001;16:312–322.
51. Whittle M. Gait Analysis: An Introduction. 4th ed. Philadelphia, PA: Butterworth Heinemann; 2007.
52. Camicioli R, Licis L. Motor impairment predicts falls in specialized Alzheimer care units. Alzheimer Dis Assoc Disord. 2004;18:214–218.
53. Karakostas T, Hsiang S, Davis B, Shenk D, Pope C, Granholm A. Dementia-Specific Gait Patterns, Irrespective of Walking Task: A Signal Detection Approach. Chicago, IL: Society for Neuroscience; 2009.
54. Karakostas T, Davis B, Maclagan M, Hsiang S. Walking and conversing: gait and speech adaptations to perturbing direct questions in dementia. Poster, 3rd Int'l Cong Gait Mental Function. Parkinson Rel Disorders. 2010;16:S1169.
55. Ripich D, Carpenter B, Ziol E. Conversational cohesion patterns in men and women with Alzheimer's disease: a longitudinal study. Int J Lang Comm Dis. 2000;35:49–64.
56. SAS Institute Inc. SAS 9.1 2000–2004. Cary, NC: SAS Institute Inc. http://www.sas.com
. Accessed May 5, 2011.
57. Sheridan P, Solomon J, Kowall N, Hausdorff J. Influence of executive function on locomotor function: Divided attention increases gait variability in Alzheimer's disease. J Am Ger Soc. 2003;51:1633–1637.
58. Petterson A, Olsson E, Wahlund L. Effect of divided attention on gait in subjects with and without cognitive impairment. J Geriatric Psychiatry Neurol. 2007;20:58–62.
59. Lajoie Y, Teasdale N, Bard C, Fleury M. Attentional demands for static and dynamic equilibrium. Exp Brain Res. 1993;97:139–144.
60. Snijders A, Verstappe C, Munneke M, Bloem B. Assessing the interplay between cognition and gait in the clinical setting. J Neural Transm. 2007;114:1315–1321.
61. Haugh M. Face in interaction. J Pragmatics. 2010;42:2073–2077.
62. Davis B, Smith M. Dementia communication in residential facilities: Intersections of training and research. In Backhaus P, ed. Communication in Elderly Care: Cross-Cultural Approaches. London/New York: Continuum; 2011;17–31.