Preprint

TalkTive: A Conversational Agent Using Backchannels to Engage Older Adults in Neurocognitive Disorders Screening

Authors:
Preprint

TalkTive: A Conversational Agent Using Backchannels to Engage Older Adults in Neurocognitive Disorders Screening

If you want to read the PDF, try requesting it from the authors.

Abstract

Conversational agents (CAs) have the great potential in mitigating the clinicians' burden in screening for neurocognitive disorders among older adults. It is important, therefore, to develop CAs that can be engaging, to elicit conversational speech input from older adult participants for supporting assessment of cognitive abilities. As an initial step, this paper presents research in developing the backchanneling ability in CAs in the form of a verbal response to engage the speaker. We analyzed 246 conversations of cognitive assessments between older adults and human assessors, and derived the categories of reactive backchannels (e.g. "hmm") and proactive backchannels (e.g. "please keep going"). This is used in the development of TalkTive, a CA which can predict both timing and form of backchanneling during cognitive assessments. The study then invited 36 older adult participants to evaluate the backchanneling feature. Results show that proactive backchanneling is more appreciated by participants than reactive backchanneling.

No file available

Request Full-text Paper PDF

To read the file of this research,
you can request a copy directly from the authors.

ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
Early diagnosis of Neurocognitive Disorder (NCD) is crucial in facilitating preventive care and timely treatment to delay further progression. This paper presents the development of a state-of-the-art automatic speech recognition (ASR) system built on the DementiaBank Pitt corpus for automatic NCD detection. Speed perturbation based audio data augmentation expanded the limited elderly speech data by four times. Large quantities of out-of-domain, non-aged adult speech were exploited by cross-domain adapting a 1000-hour LibriSpeech corpus trained LF-MMI factored TDNN system to De- mentiaBank. The variability among elderly speakers was modeled using i-Vector and learning hidden unit contributions (LHUC) based speaker adaptive training. Robust Bayesian estimation of TDNN systems and LHUC transforms were used in both cross-domain and speaker adaptation. A Transformer language model was also built to improve the final system performance. A word error rate (WER) reduction of 11.72% absolute (26.11% relative) was obtained over the baseline i-Vector adapted LF-MMI TDNN system on the evaluation data of 48 elderly speakers. The best NCD detection accuracy of 88%, comparable to that using the ground truth speech transcripts, was obtained using the textual features extracted from the final ASR system outputs.
Article
Full-text available
Wearable activity trackers can motivate older adults to engage in the recommended daily amount of physical activity (PA). However, individuals may not maintain their use of the trackers over a longer period. To investigate the attitudes of activity tracker adoption and their effects on actual PA performance, we conducted a three-month study. We gave activity trackers to 16 older adults and assessed attitudes on activity tracker adoption through a survey during the study period. We extracted participants’ PA measures, step counts, and moderate and vigorous physical activity (MVPA) times. We observed significant differences in adoption attitudes during the three different periods (χ2(2, 48) = 6.27, p < 0.05), and PA measures followed similar decreasing patterns (F(83, 1357) = 12.56, 13.94, p < 0.00001). However, the Pearson correlation analysis (r = 0.268, p = 0.284) and a Bland–Altman plot indicated a bias between two PA measures. Positive attitudes at the initial stage did not persist through the study period, and both step counts and length of MVPA time showed waning patterns in the study period. The longitudinal results from both measures demonstrated the patterns of old adults’ long-term use and adoption. Considering the accuracy of the activity tracker and older adults’ athletic ability, MVPA times are more likely to be a reliable measure of older adults’ long-term use and successful adoption of activity trackers than step counts. The results support the development of better activity tracker design guidelines that would facilitate long-term adoption among older adults.
Conference Paper
Full-text available
This paper presents our latest investigation on modeling backchannel in conversations. Motivated by a proactive backchanneling theory, we aim at developing a system which acts as a proactive listener by inserting backchannels, such as continuers and assessment, to influence speakers. Our model takes into account not only lexical and acoustic cues, but also introduces the simple and novel idea of using listener embeddings to mimic different backchanneling behaviours. Our experimental results on the Switchboard benchmark dataset reveal that acoustic cues are more important than lexical cues in this task and their combination with listener embeddings works best on both, manual transcriptions and automatically generated transcriptions.
Conference Paper
Full-text available
This paper is a literature review of 57 papers that have examined the role and impact of conversational agents (CAs) in the health domain. We note that three key themes repeatedly arose during the review: therapeutic alliance, trust, and human intervention. We also point out several areas that have been largely overlooked, such as specific patient characteristics that influence the effects of CA usage, the results of differing CA designs, and specific human-CA relationships. Based on the current gaps in scholarship, we recommend several future intersections at which CAs and healthcare can meet.
Conference Paper
Full-text available
Picture description tasks are used for the detection of cognitive decline associated with Alzheimer’s disease (AD). Recent years have seen work on automatic AD detection in picture descriptions based on acoustic and word-based analysis of the speech. These methods have shown some success but lack an ability to capture any higher level effects of cognitive decline on the patient’s language. In this paper, we propose a novel model that encompasses both the hierarchical and sequential structure of the description and detect its informative units by attention mechanism. Automatic speech recognition (ASR) and punctuation restoration are used to transcribe and segment the data. Using the DementiaBank database of people with AD as well as healthy controls (HC), we obtain an F-score of 84.43% and 74.37% when using manual and automatic transcripts respectively. We further explore the effect of adding additional data (a total of 33 descriptions collected using a ‘ digital doctor’ ) during model training, and increase the F-score when using ASR transcripts to 76.09%. This outperforms baseline models, including bidirectional LSTM and bidirectional hierarchical neural network without an attention mechanism, and demonstrate that the use of hierarchical models with attention mechanism improves the AD/HC discrimination performance.
Conference Paper
Full-text available
A series of recent studies have shed light on the existence of sociocultural inequities in collaborative learning environments. We present IneqDetect, a system which helps students reflect on the way that they communicate as a team. Conversations during collaborative learning activities are recorded using lapel microphones, processed to determine who spoke at a given time, and then visualized. The resulting dashboard visualization provides students with a timeline of when each student was speaking, a summary of how much they spoke, and an estimate of how equitable the conversation was between team members. Students reflect on this information at the end of the class period to identify and address issues, such as conversational inequality, within their groups. IneqDetect was deployed across four CS active learning classrooms. IneqDetect led students to discuss group dynamics, change their behaviors, and gain insights about themselves and their team. However, conversational equity within groups did not improve.
Conference Paper
Full-text available
In this paper we present the results of an exploratory study examining the potential of voice assistants (VA) for some groups of older adults in the context of Smart Home Technology (SHT). To research the aspect of older adults' interaction with voice user interfaces (VUI) we organized two workshops and gathered insights concerning possible benefits and barriers to the use of VA combined with SHT by older adults. Apart from evaluating the participants' interaction with the devices during the two workshops we also discuss some improvements to the VA interaction paradigm.
Article
Full-text available
Over the last decade, there has been an explosion of digital interventions that aim to either supplement or replace face-to-face mental health services. More recently, a number of automated conversational agents have also been made available, which respond to users in ways that mirror a real-life interaction. What are the social and ethical concerns that arise from these advances? In this article, we discuss, from a young person’s perspective, the strengths and limitations of using chatbots in mental health support. We also outline what we consider to be minimum ethical standards for these platforms, including issues surrounding privacy and confidentiality, efficacy, and safety, and review three existing platforms (Woebot, Joy, and Wysa) according to our proposed framework. It is our hope that this article will stimulate ethical debate among app developers, practitioners, young people, and other stakeholders, and inspire ethically responsible practice in digital mental health.
Article
Full-text available
Background Wearable activity trackers offer the opportunity to increase physical activity through continuous monitoring. Viewing tracker use as a beneficial health behavior, we explored the factors that facilitate and hinder long-term activity tracker use, applying the transtheoretical model of behavior change with the focus on the maintenance stage and relapse. Objective The aim of this study was to investigate older adults’ perceptions and uses of activity trackers at different points of use: from nonuse and short-term use to long-term use and abandoned use to determine the factors to maintain tracker use and prevent users from discontinuing tracker usage. Methods Data for the research come from 10 focus groups. Of them, 4 focus groups included participants who had never used activity trackers (n=17). These focus groups included an activity tracker trial. The other 6 focus groups (without the activity tracker trial) were conducted with short-term (n=9), long-term (n=11), and former tracker users (n=11; 2 focus groups per user type). Results The results revealed that older adults in different tracker use stages liked and wished for different tracker features, with long-term users (users in the maintenance stage) being the most diverse and sophisticated users of the technology. Long-term users had developed a habit of tracker use whereas other participants made an effort to employ various encouragement strategies to ensure behavior maintenance. Social support through collaboration was the primary motivator for long-term users to maintain activity tracker use. Short-term and former users focused on competition, and nonusers engaged in vicarious tracker use experiences. Former users, or those who relapsed by abandoning their trackers, indicated that activity tracker use was fueled by curiosity in quantifying daily physical activity rather than the desire to increase physical activity. Long-term users saw a greater range of pros in activity tracker use whereas others focused on the cons of this behavior. Conclusions The results suggest that activity trackers may be an effective technology to encourage physical activity among older adults, especially those who have never tried it. However, initial positive response to tracker use does not guarantee tracker use maintenance. Maintenance depends on recognizing the long-term benefits of tracker use, social support, and internal motivation. Nonadoption and relapse may occur because of technology’s limitations and gaining awareness of one’s physical activity without changing the physical activity level itself.
Article
Full-text available
Using supporting backchannel (BC) cues can make human-computer interaction more social. BCs provide a feedback from the listener to the speaker indicating to the speaker that he is still listened to. BCs can be expressed in different ways, depending on the modality of the interaction, for example as gestures or acoustic cues. In this work, we only considered acoustic cues. We are proposing an approach towards detecting BC opportunities based on acoustic input features like power and pitch. While other works in the field rely on the use of a hand-written rule set or specialized features, we made use of artificial neural networks. They are capable of deriving higher order features from input features themselves. In our setup, we first used a fully connected feed-forward network to establish an updated baseline in comparison to our previously proposed setup. We also extended this setup by the use of Long Short-Term Memory (LSTM) networks which have shown to outperform feed-forward based setups on various tasks. Our best system achieved an F1-Score of 0.37 using power and pitch features. Adding linguistic information using word2vec, the score increased to 0.39.
Conference Paper
Full-text available
While there has been a growing body of work in child-robot interaction, we still have very little knowledge regarding young children's speaking and listening dynamics and how a robot companion should decode these behaviors and encode its own in a way children can understand. In developing a backchannel prediction model based on observed nonverbal behaviors of 4-6 year-old children, we investigate the effects of an attentive listening robot on a child's storytelling. We provide an extensive analysis of young children's nonverbal behavior with respect to how they encode and decode listener responses and speaker cues. Through a collected video corpus of peer-to-peer storytelling interactions, we identify attention-related listener behaviors as well as speaker cues that prompt opportunities for listener backchannels. Based on our findings, we developed a backchannel opportunity prediction (BOP) model that detects four main speaker cue events based on prosodic features in a child's speech. This rule-based model is capable of accurately predicting backchanneling opportunities in our corpora. We further evaluate this model in a human-subjects experiment where children told stories to an audience of two robots, each with a different backchanneling strategy. We find that our BOP model produces contingent backchannel responses that conveys an increased perception of an attentive listener, and children prefer telling stories to the BOP model robot.
Article
Full-text available
Background: Although memory impairment is the main symptom of Alzheimer’s disease (AD), language impairment can be an important marker. Relatively few studies of language in AD quantify the impairments in connected speech using computational techniques. Objective: We aim to demonstrate state-of-the-art accuracy in automatically identifying Alzheimer’s disease from short narrative samples elicited with a picture description task, and to uncover the salient linguistic factors with a statistical factor analysis. Methods: Data are derived from the DementiaBank corpus, from which 167 patients diagnosed with “possible” or “probable” AD provide 240 narrative samples, and 97 controls provide an additional 233. We compute a number of linguistic variables from the transcripts, and acoustic variables from the associated audio files, and use these variables to train a machine learning classifier to distinguish between participants with AD and healthy controls. To examine the degree of heterogeneity of linguistic impairments in AD, we follow an exploratory factor analysis on these measures of speech and language with an oblique promax rotation, and provide interpretation for the resulting factors. Results: We obtain state-of-the-art classification accuracies of over 81% in distinguishing individuals with AD from those without based on short samples of their language on a picture description task. Four clear factors emerge: semantic impairment, acoustic abnormality, syntactic impairment, and information impairment. Conclusion: Modern machine learning and linguistic analysis will be increasingly useful in assessment and clustering of suspected AD.
Article
Full-text available
Introduction: Among the nonmotor features of Parkinson's disease (PD), cognitive impairment is one of the most troublesome problems. New diagnostic criteria for mild and major neurocognitive disorder (NCD) in PD were established by Diagnostic and Statistical Manual of Mental Disorders 5th edition (DSM-5). The aim of our study was to establish the diagnostic accuracy of widely used screening tests for NCD in PD. Methods: Within the scope of our study we evaluated the sensitivity and specificity of different neuropsychological tests (Addenbrooke's Cognitive Examination (ACE), Mattis Dementia Rating Scale (MDRS), Mini Mental State Examination (MMSE), and Montreal Cognitive Assessment (MoCA)) in 370 PD patients without depression. Results: MoCA and ACE feature the finest diagnostic accuracy for detecting mild cognitive disorder in PD (DSM-5) at the cut-off scores of 23.5 and 83.5 points, respectively. The diagnostic accuracy of these tests was 0.859 (95% CI: 0.818-0.894, MoCA) and 0.820 (95% CI: 0.774-0.859, ACE). In the detection of major NCD (DSM-5), MoCA and MDRS tests exhibited the best diagnostic accuracy at the cut-off scores of 20.5 and 132.5 points, respectively. The diagnostic accuracy of these tests was 0.863 (95% CI: 0.823-0.897, MoCA) and 0.830 (95% CI: 0.785-0.869, MDRS). Conclusion: Our study demonstrated that the MoCA may be the most suitable test for detecting mild and major NCD in PD.
Conference Paper
Full-text available
Smart homes have significant potential to enhance the lives of older adults, extending the period of healthy ageing, through monitoring wellbeing, detecting decline and applying interventions to prevent or slow down this decline. In this paper we present results from interviews with 7 older adults who have been living in smart homes for over 4 years. Our aims were to 1) examine attitudes to living with sensors and AAL technology over time; 2) gather opinions on the usefulness of this data for supporting self-management of health and wellbeing and 3) evaluate the effectiveness of various visualization techniques for presenting sensor-based health and wellness data. Our findings show that older adults are interested in receiving feedback from sensor technology to support them self-managing their wellbeing. Potential beneficial information includes time spent inside and outside the home, walking time, sleep, activity, blood pressure and weight. This information needs to be enhanced by education and goal-setting and by representing data using visualisations that are simple and intuitive.
Article
Full-text available
Ageing has become a significant area of interest in Human-Computer Interaction (HCI) in recent years. In this article we provide a critical analysis of 30 years of ageing research published across the ACM Special Interest Group on Computer-Human Interaction (SIGCHI) community. Discourse analysis of the content of 644 archival papers highlights how ageing is typically framed as a “problem” that can be managed by technology. We highlight how ageing is typically defined through an emphasis on the economic and societal impact of health and care needs of older people, concerns around socialisation as people age, and declines in abilities and associated reductions in performance when using technology. We draw from research within the fields of social and critical gerontology to highlight how these discourses in SIGCHI literature represent common stereotypes around old age that have also prevailed in the wider literature in gerontology. We conclude by proposing strategies for future research at the intersection of ageing and HCI.
Article
Full-text available
Brief addressee responses such as uh huh, oh, and wow, which are called backchannels, are typically considered reactive phenomena – devices that respond in various ways to what was just said. Addressees, in providing backchannels, actively shape story telling in spontaneous dialogue ( Bavelas et al., 2000). We contrasted generic backchannels with context-sensitive specific backchannels within a collection of face-to-face dialogues and in a narrative completion experiment. The analysis demonstrates that storytellers respond in distinct patterns to the two categories of backchannels. After generic backchannels, they provide discourse-new events. After specific backchannels, they provide elaborative information on previously presented events. Results from an experiment support this analysis, indicating that people reading transcripts of the conversation predict a similar pattern of story continuation following generic versus specific backchannels. We conclude that addressee responses are not only reactive, but proactive and collaborative in the shaping of narrative.
Conference Paper
Full-text available
The INTERSPEECH 2013 Computational Paralinguistics Challenge provides for the first time a unified test-bed for Social Signals such as laughter in speech. It further introduces conflict in group discussions as a new task and deals with autism and its manifestations in speech. Finally, emotion is revisited as task, albeit with a broader range of overall twelve enacted emotional states. In this paper, we describe these four Sub-Challenges, their conditions, baselines, and a new feature set by the openSMILE toolkit, provided to the participants.
Article
Full-text available
The paper traces the vicissitudes of the Yerkes-Dodson law from 1908 to the present. In its original form, the law was intended to describe the relation between stimulus strength and habit-formation for tasks varying in discrimination difficultness. But later generations of investigations and textbook authors have rendered it variously as the effects of punishment, reward, motivation, drive, arousal, anxiety, tension or stress upon learning, performance, problem-solving, coping or memory; while the task variable has been commonly referred to as difficulty, complexity or novelty, when it is not omitted altogether. These changes are seldom explicitly discussed, and are often misattributed to Yerkes and Dodson themselves. The various reformulations are seen as reflecting conceptual changes and current developments in the areas of learning, motivation and emotion, and it is argued that the plasticity of the law also reflects the vagueness of basic psychological concepts in these areas.
Article
Full-text available
This study tests the hypothesis that driver stress is associated with performance impairment because stress-prone drivers are vulnerable to overload of attentional resources. Eighty young-adult subjects performed a simulated drive concurrently with a grammatical reasoning task, presented either visually or auditorily. Priority assigned to the 2 tasks was also manipulated. In general, the patterns of dual-task interference predicted by attentional resource theory were not found, although interference was apparent with the auditory reasoning task. Measures of vulnerability to driver stress and intrusive cognitions were related to impaired lateral control mainly when task demands were relatively low, contrary to the overload hypothesis. These data indicate that performance in this task paradigm is characterized by adaptive mobilization of effort to meet changing task demands. Stressed drivers adapted to high levels of demand fairly efficiently, but they may be at risk of performance impairment when the task requires relatively little active control. Advantages and disadvantages of the simulator approach are discussed.
Article
Full-text available
Face-to-face conversation is a unique listening setting, with a particular kind of listener; the person the speaker is directly addressing is the addressee. Our research program has included several experiments involving detailed, reliable examinations of the subtle yet crucial behaviors that addressees use to collaborate with the speaker in face-to-face dialogue. We have found that addressees respond to speakers using either generic back channels (e.g., “m-hm” or nodding) or responses that specify what the addressee has understood (e.g., opening eyes wide to show surprise). Addressees timed these specific responses to precise moments in the speaker's narrative, and they tailored their responses to that moment (e.g., wincing when the speaker described something painful). Distracting addressees with a task that prevented them from following the speakers' narratives made these addressees unable to contribute specific responses, which, in turn, had a deleterious effect on the speakers' storytelling. Further research showed that addressees who were not distracted used a wide variety of behaviors to contribute to dialogue without interrupting the speaker, such as brief vocalizations, facial displays, and even gestures. Speakers and addressees regulated the timing of addressee responses using an interactive pattern of gaze. Addressees also indicated understanding by their formulations, which summarized or paraphrased what the speaker had said. However, our analysis showed that these formulations were not neutral. The analysis of addressees in face-to-face dialogue generates a deeper understanding of the listening process and has implications for listening in applied settings, such as psychotherapy or health care interactions.
Conference Paper
Recent advancements and economic feasibility have led to the widespread adoption of conversational digital assistants for everyday work. While research has focused on the use of these conversational assistants such as Siri, Google Assistant or Alexa, by young adults and families, very little work focuses on the acceptance and adaptability amongst the older adults. This SIG aims to discuss the use and benefits of these conversational digital assistants for the well being of older adults. The goals for this SIG are to (i) explore the acceptance/adoption of voice-based conversational agents for older adults. (ii) explore anthropomorphism in the design of conversational digital assistants. (iii) understand triggers (scenarios of use) that can initiate the process of reminiscence thus leading to meaningful conversation. (iv) explore conversational User Experience. (v) explore the coexistence of non-conversational use cases.
Article
As voice-based conversational agents such as Amazon Alexa and Google Assistant move into our homes, researchers have studied the corresponding privacy implications, embeddedness in these complex social environments, and use by specific user groups. Yet it is unknown how users categorize these devices: are they thought of as just another object, like a toaster? As a social companion? Though past work hints to human-like attributes that are ported onto these devices, the anthropomorphization of voice assistants has not been studied in depth. Through a study deploying Amazon Echo Dot Devices in the homes of older adults, we provide a preliminary assessment of how individuals 1) perceive having social interactions with the voice agent, and 2) ontologically categorize the voice assistants. Our discussion contributes to an understanding of how well-developed theories of anthropomorphism apply to voice assistants, such as how the socioemotional context of the user (e.g., loneliness) drives increased anthropomorphism. We conclude with recommendations for designing voice assistants with the ontological category in mind, as well as implications for the design of technologies for social companionship for older adults.
Article
Neurocognitive disorders create important challenges for patients, their families, and clinicians who provide their health care. Early/timely detection in daily clinical practice allows for diagnosis and adequate treatment, psychosocial support, education, and engagement in shared decision-making related to health care, life planning, involvement in research, and financial matters. However, neurocognitive disorders, when present, are not detected or not diagnosed and not documented, in more than half of patients seen by primary care physicians. The aim of this paper is to highlight the strategies and the perspectives to improve the early/timely detection of neurocognitive disorders in daily clinical practice.
Article
Objective: The aim of this review was to explore the current evidence for conversational agents or chatbots in the field of psychiatry and their role in screening, diagnosis, and treatment of mental illnesses. Methods: A systematic literature search in June 2018 was conducted in PubMed, EmBase, PsycINFO, Cochrane, Web of Science, and IEEE Xplore. Studies were included that involved a chatbot in a mental health setting focusing on populations with or at high risk of developing depression, anxiety, schizophrenia, bipolar, and substance abuse disorders. Results: From the selected databases, 1466 records were retrieved and 8 studies met the inclusion criteria. Two additional studies were included from reference list screening for a total of 10 included studies. Overall, potential for conversational agents in psychiatric use was reported to be high across all studies. In particular, conversational agents showed potential for benefit in psychoeducation and self-adherence. In addition, satisfaction rating of chatbots was high across all studies, suggesting that they would be an effective and enjoyable tool in psychiatric treatment. Conclusion: Preliminary evidence for psychiatric use of chatbots is favourable. However, given the heterogeneity of the reviewed studies, further research with standardized outcomes reporting is required to more thoroughly examine the effectiveness of conversational agents. Regardless, early evidence shows that with the proper approach and research, the mental health field could use conversational agents in psychiatric treatment.
Article
Neurogenerative disorders, like dementia, can affect a person's speech, language and as a consequence, conversational interaction capabilities. A recent study, aimed at improving dementia detection accuracy, investigated the use of conversation analysis (CA) of interviews between patients and neurologists as a means to differentiate between patients with progressive neurodegenerative memory disorder (ND) and those with (non-progressive) functional memory disorders (FMD). However, doing manual CA is expensive and difficult to scale up for routine clinical use. In this paper, we present an automatic classification system using an intelligent virtual agent (IVA). In particular, using two parallel corpora of respectively neurologist- and IVA-led interactions, we show that using acoustic, lexical and CA-inspired features enable ND/FMD classification rates of 90.0% for the neurologist-patient conversations, and 90.9% for the IVA-patient conversations. Analysis of the differentiating potential of individual features show that some differences exist between the IVA and human-led conversations, for example in average turn length of patients.
Article
Background: Semantic verbal fluency (SVF) tests are routinely used in screening for mild cognitive impairment (MCI). In this task, participants name as many items as possible of a semantic category under a time constraint. Clinicians measure task performance manually by summing the number of correct words and errors. More fine-grained variables add valuable information to clinical assessment, but are time-consuming. Therefore, the aim of this study is to investigate whether automatic analysis of the SVF could provide these as accurate as manual and thus, support qualitative screening of neurocognitive impairment. Methods: SVF data were collected from 95 older people with MCI (n = 47), Alzheimer's or related dementias (ADRD; n = 24), and healthy controls (HC; n = 24). All data were annotated manually and automatically with clusters and switches. The obtained metrics were validated using a classifier to distinguish HC, MCI, and ADRD. Results: Automatically extracted clusters and switches were highly correlated (r = 0.9) with manually established values, and performed as well on the classification task separating HC from persons with ADRD (area under curve [AUC] = 0.939) and MCI (AUC = 0.758). Conclusion: The results show that it is possible to automate fine-grained analyses of SVF data for the assessment of cognitive decline.
Conference Paper
This paper explores how seniors perceive Voice Enabled User Interfaces (VUIs) and the factors that shape those perceptions. An experiment was administered to 15 seniors (over age of 65), in which the participants searched for information using a traditional keyboard/mouse interface and an experimental voice/touch interface. An analysis of the data collected showed that seniors perceive meaningful differences between the two interfaces in terms of learnability, usability, ease of understanding and helpfulness.
Article
When considering the future and application of wearable technological devices, the growing demographic of adults over the age of 65 should be included during the product design process. Wearable technology could provide an ideal mechanism for delivering supportive applications to a growing older adult population, but designers must consider age-related changes in cognitive, sensory, and motor function when developing for older populations. Potential issues with wearable devices for older adults can be avoided by acknowledging limitations, and development teams can create effective and safe platforms that appeal to a variety of end users.
Article
This article provides information for Extension professionals on the correct analysis of Likert data. The analyses of Likert-type and Likert scale data require unique data analysis procedures, and as a result, misuses and/or mistakes often occur. This article discusses the differences between Likert-type and Likert scale data and provides recommendations for descriptive statistics to be used during the analysis. Once a researcher understands the difference between Likert-type and Likert scale data, the decision on appropriate statistical procedures will be apparent.
Article
Empathy is considered a key curative aspect of interactive counseling based psychotherapy. In this present work, it is deemed an interpersonal behavior whereby one person communicates attention and understanding to another. The process of empathy involves "trying on" the thoughts or feelings of another person. Thus it is hypothesized to involve entrainment, wherein interlocutors become more alike in behaviors such as speech, gestures, emotions, etc. We extend previous algorithms on vocal similarity, measured through temporal weighting on speech features, and principal components constructed from these features. In addition, we approximate entrainment via turn-based differences in weighted pitch between speakers and turn-taking statistics. Results show these cues are significantly correlated with human ratings of empathy, and can predict therapist empathy significantly better than chance. This work establishes a link between empathy and entrainment, and proposes computational approaches to infer therapist empathy.
Article
Stability Selection was recently introduced by Meinshausen and B¨uhlmann (2010) as a very general technique designed to improve the performance of a variable selection algorithm. It is based on aggregating the results of applying a selection procedure to subsamples of the data. We introduce a variant, called Complementary Pairs Stability Selection (CPSS), and derive bounds both on the expected number of variables included by CPSS that have low selection probability under the original procedure, and on the expected number of high selection probability variables that are excluded. These results require no (e.g. exchangeability) assumptions on the underlying model or on the quality of the original selection procedure. Under reasonable shape restrictions, the bounds can be further tightened, yielding improved error control, and therefore increasing the applicability of the methodology.
Article
In recent years, researchers have tried to create unhindered human-computer interaction by giving virtual agents human-like conversational skills. Predicting backchannel feedback for agent listeners has become a novel research hot-spot. The main goal of this paper is to identify appropriate features and methods for backchannel prediction in Mandarin conversations. Firstly, multimodal Mandarin conversations are recorded for the analysis of backchannel behaviors. In order to eliminate individual difference in the original face-to-face conversations, more backchannels from different listeners are gathered together. These data confirm that backchannels occurring in the speakers&apos; pauses form a vast majority in Mandarin conversations. Both prosodic and visual features are used in backchannel prediction. Four types of models based on the speakers&apos; pauses are built by using support vector machine classifiers. An evaluation of the pause-based prediction model has shown relatively high accuracy in consideration of the optional nature of backchannel feedback. Finally, the results of the subjective evaluation validate that the conversations performed between humans and virtual listeners using backchannels predicted by the proposed models is more unhindered compared to other backchannel prediction methods.
Article
Listenership (consisting of backchannel feedback) and its effect on intercultural communication were investigated in 30 dyadic conversations in English between Japanese and American participants. The findings of this study demonstrate several differences in how members of each culture used backchannels in terms of frequency, variability, placement, and function. This study also found evidence supporting the hypothesis that backchannel conventions that are not shared between cultures contribute to negative perceptions across cultures. Thus, the findings of this study support the conclusion that listenership warrants more attention in EFL classes in Japan. Further, toward creating a pedagogical framework, this study also provides EFL professionals with a comprehensive account of the listenership of Japanese EFL speakers. Finally, this study also offers potential insights into the linguistic variation of native English speakers. That is, the negative perceptions that the American native English speakers reported of their Japanese EFL speaking interlocutors' listener responses in this study were not as pronounced as those reported by the British native English speakers in a previous study conducted by the same researcher.
Article
Evaluation constitutes a central feature of personal stories in conversation. Storytellers introduce evaluation into their narratives in various ways, including cases of appropriating assessments offered by their listeners. A storyteller may orient to the content of listener assessments and respond to them in various (positive or negative) ways, suspending the narrative in progress to comment or altering its direction. Shared assessments can lead to higher involvement and increased rapport with consequences for subsequent interaction between the participants. Rejections of listener assessments are much less frequent than ratifications: rejection of a listener assessment expresses the teller's refusal to have it count as part of the overall evaluation of the story in progress.
Article
Mobile computing devices, such as smart phones, offer benefits that may be especially valuable to older adults (age 65+). Yet, older adults have been shown to have difficulty learning to use these devices. In the research presented in this article, we sought to better understand how older adults learn to use mobile devices, their preferences and barriers, in order to find new ways to support them in their learning process. We conducted two complementary studies: a survey study with 131 respondents from three age groups (20--49, 50--64, 65+) and an in-depth field study with 6 older adults aged 50+. The results showed, among other things, that the preference for trial-and-error decreases with age, and while over half of older respondents and participants preferred using the instruction manual, many reported difficulties using it. We discuss implications for design and illustrate these implications with an example help system, Help Kiosk, designed to support older adults’ learning to use mobile devices.
Conference Paper
It is recognized that empowering individuals to manage their own health and wellbeing will result in more cost-effective healthcare systems, improved health outcomes and will encourage healthy individuals to remain that way. With the advent of the quantifiedself movement in recent years, there has been an increase in technology applications supporting wellness self-management. Such applications allow people to self-track and self-report, with many providing feedback. However, little research in this area has examined how best to support older adults in health selfmanagement. This paper reports findings from a 5-month home deployment of YourWellness - an application that supports older adults in self-reporting on their wellbeing and provides feedback to promote positive wellbeing management. Our findings contribute to a greater understanding of older adults' attitudes and behaviours in relation to wellbeing self-management that can facilitate the creation of new, personalized health and wellbeing interventions for this population.
Article
This article describes listening practices in English conversation from an Emancipatory Pragmatics perspective, focusing on the role of the listener as a modality of action and seeking to evaluate linguistic behaviors like responses in terms of cultural assumptions about politeness, turn-taking, silence, and overlapping talk. In producing minimal response tokens, a listener signals a willingness to remain (predominantly) silent, to refrain from interrupting and to attend to the primary speaker, and thereby encourages the speaker to continue with a multi-unit turn. But even single word responses can have a significant effect on the trajectory of an extended turn by another speaker. Response tokens differ widely in their degree of obtrusiveness, such that some listener responses like uh-huh attract little or no attention to themselves and essentially never evoke a specific response of their own, while assessments like wow, on through signals of processing difficulty like oh, and challenges like so increasingly attract the attention of the primary speaker and elicit a response in their own right. This ranking in terms of obtrusiveness or insistency differs from other sub-classifications or scales so far described in the literature on listener responses.
Article
This article describes the listening practices of interviewers in US American television celebrity interviews in the framework of a more general account of listener activities. It explores how interviewers signal listenership, emotional involvement, and the uptake of information, how they prompt, aid and act as a foil to interviewees, and how these practices may affect the audience and the trajectory of the interview in progress. The purview moves from interviewer activities which match practices typical of listeners in everyday conversation to strategies directed at guiding or entertaining the audience. Interviewer responses may be directed primarily at the audience rather than the interviewee, and they can be more or less obtrusive or manipulative. Interviews are investigated in which Oprah Winfrey and Larry King go so far as to answer their own questions and to engage in co-narration and in the construction of direct speech on behalf of their interviewees. Jay Leno collaborates with a guest to produce an example of a team performance oriented toward humor built around the traditional interview situation.