Search

Background: This article uncovers why people with severe expressive aphasia’s turns-at-talk are sometimes not treated as producing an action by their communication partners, and the impact this has on the person with aphasia’s (PWA’s) agency. We demonstrate resources PWAs use to pursue talk and which assist with the production of a recognizable action. Method: We examined turns produced by four PWAs and their communication partners (CPs), where present, using conversation analysis, identifying features that do not receive a response and features promoting action ascription. Analysis: The PWAs’ semantically empty or unclear turns, turns lacking sequential context, or the CPs’ focus on their own actions led to a lack of action ascription. However, CPs do attend to PWAs’ multimodal features of interaction, and PWAs’ repetition accompanied by an upgraded gesture was shown to pursue a response. Action ascription was aided by the PWAs’ preserved use of silence as a communicative device. Discussion: When PWAs’ actions are not appropriately ascribed, their agency may be diminished. Communication partners should attend to all features of the PWA’s turns, including gesture and silence, to progress the PWA’s action, rather than their own misappropriated action. This may mean accepting a delay in progressivity while the PWA pursues an appropriate response. Through this, the PWA’s agency in interaction can be maintained, and intersubjectivity achieved.

Objective\ud \ud There are three common causes of Transient Loss of Consciousness (TLOC), syncope, epileptic and psychogenic nonepileptic seizures (PNES). Many individuals who have experienced TLOC initially receive an incorrect diagnosis and inappropriate treatment. Whereas syncope can be distinguished relatively easily with a small number of “yes”/”no” questions, the differentiation of the other two causes of TLOC is more challenging. Previous qualitative research based on the methodology of Conversation Analysis has demonstrated that the descriptions of epileptic seizures contain more formulation effort than accounts of PNES. This research investigates whether features likely to reflect the level of formulation effort can be automatically elicited from audio recordings and transcripts of speech and used to differentiate between epileptic and nonepileptic seizures.\ud \ud \ud Method\ud \ud Verbatim transcripts of conversations between patients and neurologists were manually produced from video and audio recordings of 45 interactions (21 epilepsy and 24 PNES). The subsection of each transcript containing the person's account of their first seizure was manually extracted for the analysis. Seven automatically detectable features were designed as markers of formulation effort. These features were used to train a Random Forest machine learning classifier.\ud \ud \ud Result\ud \ud There were significantly more hesitations and repetitions in descriptions of epileptic than nonepileptic seizures. Using a nested leave-one-out cross validation approach, 71% of seizures were correctly classified by the Random Forest classifier.\ud \ud \ud Discussion\ud \ud This pilot study provides proof of principle that linguistic features that have been automatically extracted from audio recordings and transcripts could be used to distinguish between epileptic seizures and PNES and thereby contribute to the differential diagnosis of TLOC. Future research should explore whether additional observations can be incorporated into a diagnostic stratification tool and compare the performance of these features when they are combined with additional information provided by patients and witnesses about seizure manifestations and medical history.

IntroductionRecent years have seen an almost sevenfold rise in referrals to specialist memory clinics. This has been associated with an increased proportion of patients referred with functional cognitive disorder (FCD), that is, non-progressive cognitive complaints. These patients are likely to benefit from a range of interventions (eg, psychotherapy) distinct from the requirements of patients with neurodegenerative cognitive disorders. We have developed a fully automated system, ‘CognoSpeak’, which enables risk stratification at the primary–secondary care interface and ongoing monitoring of patients with memory concerns.MethodsWe recruited 15 participants to each of four groups: Alzheimer’s disease (AD), mild cognitive impairment (MCI), FCD and healthy controls. Participants responded to 12 questions posed by a computer-presented talking head. Automatic analysis of the audio and speech data involved speaker segmentation, automatic speech recognition and machine learning classification.ResultsCognoSpeak could distinguish between participants in the AD or MCI groups and those in the FCD or healthy control groups with a sensitivity of 86.7%. Patients with MCI were identified with a sensitivity of 80%.DiscussionOur fully automated system achieved levels of accuracy comparable to currently available, manually administered assessments. Greater accuracy should be achievable through further system training with a greater number of users, the inclusion of verbal fluency tasks and repeat assessments. The current data supports CognoSpeak’s promise as a screening and monitoring tool for patients with MCI. Pending confirmation of these findings, it may allow clinicians to offer patients at low risk of dementia earlier reassurance and relieve pressures on specialist memory services.

Speech-based automatic approaches for detecting neurodegenerative disorders (ND) and mild cognitive impairment (MCI) have received more attention recently due to being non-invasive and potentially more sensitive than current pen-and-paper tests. The performance of such systems is highly dependent on the choice of features in the classification pipeline. In particular for acoustic features, arriving at a consensus for a best feature set has proven challenging. This paper explores using deep neural network for extracting features directly from the speech signal as a solution to this. Compared with hand-crafted features, more information is present in the raw waveform, but the feature extraction process becomes more complex and less interpretable which is often undesirable in medical domains. Using a SincNet as a first layer allows for some analysis of learned features. We propose and evaluate the Sinc-CLA (with SincNet, Convolutional, Long Short-Term Memory and Attention layers) as a task-driven acoustic feature extractor for classifying MCI, ND and healthy controls (HC). Experiments are carried out on an in-house dataset. Compared with the popular hand-crafted feature sets, the learned task-driven features achieve a superior classification accuracy. The filters of the SincNet is inspected and acoustic differences between HC, MCI and ND are found.

A body of research has shown that there are linguistic differences in the way people with epilepsy talk about their seizures when compared to those with non-epileptic seizures. We extend this line of research by presenting the results of a phonetic analysis comparing speech samples from people with a confirmed diagnosis of epilepsy (7 patients), to those with a confirmed diagnosis of non-epileptic seizures (8 patients). Variables considered include features of pitch, intensity, duration and pausing in their responses to questions from a neurologist during medical history-taking. We find only limited evidence of differences between the two diagnostic groups (epilepsy vs. non-epileptic seizures). We discuss possible reasons for this lack of evidence.

This study examines the linguistic and interactional organization of repair in Brazilian Portuguese playtime conversations between six mothers and their children (mean age 2;6). Following both interactional phonetics and conversation analytic methodological approaches, this investigation focuses on how children and mothers negotiate the action done by the mother's lexical repetition used to initiate repair on the child's previous turn. The results suggest that children's ability to understand mothers' lexical repetitions addressing pronunciation problems comes before their ability to understand repetitions that address problems of lexical choice.

This study uses the tools of Conversation Analysis (CA) to investigate problems that occur in LADO (Language Analysis for the Determination of Origin) interviews. We analysed five recorded interviews with female asylum seekers, focussing on question and response pairs. Several problems were identified, associated with directives, echo questions, and challenges. The study also looked at how repair is initiated and carried out. Directives were frequently issued as part of multiple questions from the interviewer, alongside additional questions or modifiers. Interviewees typically provided an answer to the most specific and/or most recent question rather than fulfilling the directive itself. Directives were also used to elicit language samples, and it was found that including a clear topic for talk was the most effective way of accomplishing this goal. Echo questions were predominantly used for requesting confirmation, and were occasionally interpreted as performing this function even where there was evidence that interviewers were using echo questions to prompt for more information or to initiate repair. Challenges contributed to a hostile atmosphere in interviews. Similarly, repair prefaced by initial but was found to be potentially hostile in some instances. Various modes of accomplishing repair were also investigated, but their effectiveness was variable. In assessing the set of question and response pairs in the recordings, we make a number of practical recommendations for improving interview practice in LADO.

Background Aphasia assessment is traditionally divided into formal and informal approaches. Informal assessment is useful in developing a rich understanding of the person with aphasia's performance, e.g., describing performance in the context of real-world activities, and exploring the impact of environmental and/or partner supports upon communication. However, defining ‘informal assessment’ is problematic and can result in clinical issues including idiosyncratic practices regarding why, when and how to apply informal assessment. Aims To examine the extent to which the informal assessment literature can guide speech and language therapists (SLTs) in their clinical application of informal assessment for post-stroke aphasia. Methods & Procedures A scoping review methodology was used. A systematic search of electronic databases (Scopus, Embase, PyscInfo, CINAHL, Ovid Medline and AMED) gave informal assessment references between 2000 and 2017 to which title/abstract and full-text screening against inclusion criteria were applied. Data were extracted from 28 resulting documents using an extraction template with fields based on the review's purpose. Main Contribution This review examines the informal assessment guidance regarding: rationale; areas of interest for informal assessment; available methods; procedural guidance; documentation; and analytical frameworks. The rationale for using informal assessment included several aspects such as gaining a ‘representative’ sample of the individual's language. Ten communication areas of interest were found with 13 different assessment methods. The procedural guidance for these methods varied considerably, with the exception of conversation and semi-structured interviewing. Overall, documentation guidance was limited but numerous analytical frameworks were found. Conclusions Several informal assessment methods are available to SLTs. However, information is mixed regarding when they might be used or how they might be applied in terms of their administration, documentation and analysis.

This article analyzes two different repair initiation practices that both utilize other-repetition. We call these framing and prefacing other-repetitions and show that they are treated as m...

BACKGROUND: The early diagnosis of dementia is of great clinical and social importance. A recent study using the qualitative methodology of conversation analysis (CA) demonstrated that language and communication problems are evident during interactions between patients and neurologists, and that interactional observations can be used to differentiate between cognitive difficulties due to neurodegenerative disorders (ND) or functional memory disorders (FMD). OBJECTIVE: This study explores whether the differential diagnostic analysis of doctor-patient interactions in a memory clinic can be automated. METHODS: Verbatim transcripts of conversations between neurologists and patients initially presenting with memory problems to a specialist clinic were produced manually (15 with FMD, and 15 with ND). A range of automatically detectable features focusing on acoustic, lexical, semantic, and visual information contained in the transcripts were defined aiming to replicate the diagnostic qualitative observations. The features were used to train a set of five machine learning classifiers to distinguish between ND and FMD. RESULTS: The mean rate of correct classification between ND and FMD was 93% ranging from 97% by the Perceptron classifier to 90% by the Random Forest classifier.Using only the ten best features, the mean correct classification score increased to 95%. CONCLUSION: This pilot study provides proof-of-principle that a machine learning approach to analyzing transcripts of interactions between neurologists and patients describing memory problems can distinguish people with neurodegenerative dementia from people with FMD.

The ageing population has caused a marked increased in the number of people with cognitive decline linked with dementia. Thus, current diagnostic services are overstretched, and there is an urgent need for automating parts of the assessment process. In previous work, we demonstrated how a stratification tool built around an Intelligent Virtual Agent (IVA) eliciting a conversation by asking memory-probing questions, was able to accurately distinguish between people with a neuro-degenerative disorder (ND) and a functional memory disorder (FMD). In this paper, we extend the number of diagnostic classes to include healthy elderly controls (HCs) as well as people with mild cognitive impairment (MCI). We also investigate whether the IVA may be used for administering more standard cognitive tests, like the verbal fluency tests. A four-way classifier trained on an extended feature set achieved 48% accuracy, which improved to 62% by using just the 22 most significant features (ROC-AUC: 82%).

BackgroundThere are three common causes of Transient Loss of Consciousness (TLOC), syncope, epileptic and psychogenic nonepileptic seizures (PNES). Many individuals who have experienced TLOC initially receive an incorrect diagnosis and inappropriate treatment. Whereas syncope can be distinguished from the other two causes relatively easily with a small number of yes/no questions, the differentiation of the other two causes of TLOC is more challenging. Previous qualitative research based on the methodology of Conversation Analysis has demonstrated that epileptic and nonepileptic seizures are described differently when patients talk to clinicians about their TLOC experiences. One particularly prominent difference is that epileptic seizure descriptions are characterised by more formulation effort than accounts of nonepileptic seizures.AimThis research investigates whether features likely to reflect the level of formulation effort can be automatically elicited from audio recordings and transcripts of speech and used to differentiate between epileptic and nonepileptic seizures.MethodVerbatim transcripts of conversations between patients and neurologists were manually produced from video and audio recordings of interactions with 45 patients (21 epilepsy and24 PNES). The subsection of each transcript containing the patients account of their first seizure was manually extracted for the analysis. Seven automatically detectable features were designed as markers of formulation effort. These features were used to train a Random Forest machine learning classifier.ResultsThere were significantly more hesitations and repetitions in descriptions of first epileptic than nonepileptic seizures. Using a nested leave-one-out cross validation approach, 71% of seizures were correctly classified by the Random Forest classifier.ConclusionsThis pilot study provides proof of principle that linguistic features that have been automatically extracted from audio recordings and transcripts could be used to distinguish between epileptic seizures and PNES and thereby contribute to the differential diagnosis of TLOC. Future research should explore whether additional observations can be incorporated into a diagnostic stratification tool. Moreover, future research should explore the performance of these features when they have been extracted from transcripts produced by automatic speech recognition and when they are combined with additional information provided by patients and witnesses about seizure manifestations and medical history.

Objectives/AimsWe used our automated cognitive assessment tool to explore whether responses to questions probing recent and remote memory could aid in distinguishing between patients with early neurodegenerative disorders and those with Functional Cognitive Disorders (FCD).Hypotheses: pwFCD would have no significant differences in pause to speech ratio and measures of linguistic complexity compared to healthy controls. pwFCD would have significant differences in pause to speech ratio and measures of linguistic complexity compared to pwMCI and pwAD.MethodsWe recruited 15 participants with FCD, MCI and AD each as well as 15 healthy controls. Participants answered 12 questions posed by the ‘Digital Doctor’. Automatic processing of the audio-recorded answers involved automatic speech recognition including detecting length of pauses. Two questions probe recent memory, exploring knowledge of current affairs. Two probe remote memory, asking for autobiographical details.We analysed the data using: Pause to speech time ratio. Moving average type token ratio (MATTR): An automated measure of vocabulary richness. Computerised propositional idea density rater (CPIDR): An automated measure of propositional idea density.ResultsThere was a significant difference in the pause to speech ratio for recent memory questions for HC versus AD (P=0.0012) and MCI (pConclusionsThis study rejects both hypotheses. However, the data supports the application of linguistic measures to recent and remote memory questions in distinguishing those with MCI & AD from HC’s. Further work will investigate the utility of incorporating additional measures of lexical and grammatical complexity (word frequency, sentence structure). Longitudinal study will provide insights into which features may predict stability in FCD and HC’s and progression from MCI to AD, supporting the system’s promise as a monitoring tool.

Previous work on interactions in the memory clinic has shown that conversation analysis can be used to differentiate neurodegenerative dementia from functional memory disorder. Based on this work, a screening system was developed that uses a computerised ‘talking head’ (intelligent virtual agent) and a combination of automatic speech recognition and conversation analysis-informed programming. This system can reliably differentiate patients with functional memory disorder from those with neurodegenerative dementia by analysing the way they respond to questions from either a human doctor or the intelligent virtual agent. However, much of this computerised analysis has relied on simplistic, nonlinguistic phonetic features such as the length of pauses between talk by the two parties. To gain confidence in automation of the stratification procedure, this paper investigates whether the patients’ responses to questions asked by the intelligent virtual agent are qualitatively similar to those given in response to a doctor. All the participants in this study have a clear functional memory disorder or neurodegenerative dementia diagnosis. Analyses of patients’ responses to the intelligent virtual agent showed similar, diagnostically relevant sequential features to those found in responses to doctors’ questions. However, since the intelligent virtual agent’s questions are invariant, its use results in more consistent responses across people – regardless of diagnosis – which facilitates automatic speech recognition and makes it easier for a machine to learn patterns. Our analysis also shows why doctors do not always ask the same question in the exact same way to different patients. This sensitivity and adaptation to nuances of conversation may be interactionally helpful; for instance, altering a question may make it easier for patients to understand. While we demonstrate that some of what is said in such interactions is bound to be constructed collaboratively between doctor and patient, doctors could consider ensuring that certain, particularly important and/or relevant questions are asked in as invariant a form as possible to be better able to identify diagnostically relevant differences in patients’ responses.

Background: There is scope for additional research into the specific linguistic and sequential structures used in speech and language therapist (SLT)-led therapeutic conversations with people with aphasia (PWA). Whilst there is some evidence that SLTs use different conversational strategies than the partners of PWA, research to date has focussed mainly on measuring the effects of conversation-based therapies—not on analysing therapeutic conversations taking place between SLTs and PWA.\ud \ud Aims: This paper presents an analysis of the use of oh-prefacing by some PWA during therapeutic supported conversations with SLTs.\ud \ud Methods & Procedures: Normally occurring therapeutic conversations between SLTs and PWA after stroke were qualitatively analysed using Conversation Analysis. Interactions with five PWA were video-recorded, involving three different specialist stroke SLTs.\ud \ud Outcomes & Results: The analysis revealed a difference in the way some PWA use turns that display understanding (e.g., oh right) versus those that continue the conversation, merely claiming understanding (e.g., right). This use of oh-prefacing is similar to that described in the literature on typical conversations. In our data, SLTs are shown to treat oh-prefaced turns differently from non-oh-prefaced turns, by pursuing the topic in the latter, and progressing on to a new topic in the former.\ud \ud Conclusions: At least some PWA use oh-prefacing in the same way as non-language-impaired adults to display understanding of information versus merely claiming to understand. The SLTs in our data are shown to treat non-oh-prefaced turns as mere claims of understanding by providing the PWA with additional information, using supported conversation techniques, and pursuing additional same-topic talk, whereas oh-prefaced turns are treated as displays of understanding by being confirmed, and leading to changes of topic. This study is a first step in providing SLTs with a clearer understanding of the ways in which they are assessing the understanding of PWA, which may in turn help them better support non-therapy staff.

Aphasia assessment is traditionally divided into formal and informal approaches. Informal assessment is useful in developing a rich understanding of the person with aphasia's performance, e.g., describing performance in the context of real-world activities, and exploring the impact of environmental and/or partner supports upon communication. However, defining 'informal assessment' is problematic and can result in clinical issues including idiosyncratic practices regarding why, when and how to apply informal assessment.To examine the extent to which the informal assessment literature can guide speech and language therapists (SLTs) in their clinical application of informal assessment for post-stroke aphasia.A scoping review methodology was used. A systematic search of electronic databases (Scopus, Embase, PyscInfo, CINAHL, Ovid Medline and AMED) gave informal assessment references between 2000 and 2017 to which title/abstract and full-text screening against inclusion criteria were applied. Data were extracted from 28 resulting documents using an extraction template with fields based on the review's purpose.This review examines the informal assessment guidance regarding: rationale; areas of interest for informal assessment; available methods; procedural guidance; documentation; and analytical frameworks. The rationale for using informal assessment included several aspects such as gaining a 'representative' sample of the individual's language. Ten communication areas of interest were found with 13 different assessment methods. The procedural guidance for these methods varied considerably, with the exception of conversation and semi-structured interviewing. Overall, documentation guidance was limited but numerous analytical frameworks were found.Several informal assessment methods are available to SLTs. However, information is mixed regarding when they might be used or how they might be applied in terms of their administration, documentation and analysis.

This paper presents work on developing an automatic dementia screening test based on patients’ ability to interact and communicate — a highly cognitively demanding process where early signs of dementia can often be detected. Such a test would help general practitioners, with no specialist knowledge, make better diagnostic decisions as current tests lack specificity and sensitivity. We investigate the feasibility of basing the test on conversations between a ‘talking head’ (avatar) and a patient and we present a system for analysing such conversations for signs of dementia in the patient’s speech and language. Previously we proposed a semi-automatic system that transcribed conversations between patients and neurologists and extracted conversation analysis style features in order to differentiate between patients with progressive neurodegenerative dementia (ND) and functional memory disorders (FMD). Determining who talks when in the conversations was performed manually. In this study, we investigate a fully automatic system including speaker diarisation, and the use of additional acoustic and lexical features. Initial results from a pilot study are presented which shows that the avatar conversations can successfully classify ND/FMD with around 91% accuracy, which is in line with previous results for conversations that were led by a neurologist.

This article examines one of the ways in which matters of truth, appropriateness, and acceptability are raised and managed within the course of everyday conversation. Using the methodology of conversation analysis, we show that by repeating what another participant has said and doing so with a high rise-fall intonation contour, a speaker claims that the repeated talk is wrong and in need of correction. There is an incongruity between two versions of the worldthe one presented in the repeated speaker's talk and the one the repeating speaker knows or believes to be true, appropriate, or acceptable. The ensuing sequences are routinely expanded and morally charged as the participants jostle for epistemic or moral authority over the matter at hand and work to repair the incongruity (even if, in the end, they agree to disagree).

A recent study using Conversation Analysis (CA) has demonstrated that communication problems may be picked up during conversations between patients and neurologists, and that this can be used to differentiate between patients with (progressive neurodegenerative dementia) ND and those with (nonprogressive) functional memory disorders (FMD). This paper presents a novel automatic method for transcribing such conversations and extracting CA-style features. A range of acoustic, syntactic, semantic and visual features were automatically extracted and used to train a set of classifiers. In a proof-of-principle style study, using data recording during real neurologist-patient consultations, we demonstrate that automatically extracting CA-style features gives a classification accuracy of 95%when using verbatim transcripts. Replacing those transcripts with automatic speech recognition transcripts, we obtain a classification accuracy of 79% which improves to 90% when feature selection is applied. This is a first and encouraging step towards replacing inaccurate, potentially stressful cognitive tests with a test based on monitoring conversation capabilities that could be conducted in e.g. the privacy of the patient’s own home.

This report, arising from a study of affiliation and disaffiliation in interaction, addresses an apparently ‘anomalous’ finding in relation to complaint sequences in conversation. In some of the cases we collected in which one speaker was complaining on behalf of the other (their co-participant), taking her side in some matter, the one on whose behalf the other was complaining did not affiliate with the complaint. Instead they resisted the complaint (again, one made on their behalf) and demurred to ‘go so far’. This finding is anomalous in the sense that if A is complaining on behalf of B, in respect of some harm done to B, then it might be expected that B would go along with the complaint, and affiliate with A. To account for how it might come about that B demurs from ‘going as far as’ A, we explore how complaints are frequently introduced in conversation. We show that complaints may emerge through a progression in which ‘the complainant’ does not initially go on record with a complaint, but instead secures the other's participation in co-constructing the complaint. Hence the ‘complaint recipient’ may be the first to make the complaint explicit, in a sequence of escalating affiliation. In the ‘anomalous’ cases, it appears that this escalation goes too far for the putative complainant (B).

Objects and Aims Referrals to secondary care memory clinic has more than tripled. This has led to increased demand on diagnostic services. Conversation Analysis (CA) can help in the screening for dementia. In CA important features of the conversation between doctor, patient are analysed. In this project we created an avatar to automatically initiate conversation. Methods The avatar, a computerised head asked questions created by consulting neurologists to which patients responded verbally. Audio and video data was collected and a simple interface with two buttons (‘start/forward’ and ‘repeat’) on a keyboard was used. Audio recordings were transcribed and annotated by the CA expert. Results 24 participants were recruited. All finished the trial without problems. We received feedback from participants, many saying they felt the avatar was easy to talk to. The CA expert correctly classified in 6 of 7 cases of Functional Memory disorder (FMD) and 4 out of 6 cases of neurodegenerative dementia. Conclusion An Avatar has potential to be a low cost addition to memory screening

This article argues for the importance of describing form independently of function, especially for prosodic and phonetic forms. Form and function are often conflated by language-in-interaction researchers when they give descriptive labels to the sound of talk (e.g., “upgraded” pitch, “continuing” intonation), and that tempts researchers to see a given form as having a given function or practice—often one that is influenced by the descriptive label. I argue that we should discipline ourselves to keeping to a purely technical description of any form (practice); that will then make it possible unambiguously to show how that form contributes to a particular function (action), without presuming the relationship to be exclusive. Data are in American and British English.

In this response to Peter Auer's commentary, I revisit the question of phonetic form and interactional meaning as well as the question of what the aim of transcription actually is (or should be). What I advocate is a careful look at the ways in which our analyses link linguistic forms with actions.

Purpose To extend our previous research demonstrating that linguistic/interactional features in patients' talk can assist the challenging differential diagnosis of epilepsy and psychogenic nonepileptic seizures (PNES) by exploring the differential diagnostic potential of references to non co-present persons (third parties). Method Initial encounters were recorded between 20 seizure patients (13 with PNES, seven with epilepsy) who were subsequently diagnosed by the recording of typical seizures with video-EEG. An analyst blinded to the medical diagnoses coded and analysed transcripts. Results There were no significant differences between the two diagnostic groups in terms of the total number of third party references or references made spontaneously by patients without prompting from the doctor. However, patients with PNES made significantly more prompted references to third parties ( p =0.022). ‘Castrophising' third party references were made in 12/13 (92.3%) of encounters with PNES patients and 1/7 (14.3%) of encounters with epilepsy patients ( p =0.001, OR 72, 95% CI=3.8–1361.9). Normalising references were identified in 2/13 (15.4%) of encounters in the PNES and 6/7 (85.7%) of encounters in the epilepsy groups ( p =0.004, OR 33, 95% CI=2.5–443.6). Conclusion There are significant differences in how patients with epilepsy or patients with PNES refer to third parties. Patients with PNES are more likely to be prompted to tell doctors what others have told them about their seizures. Patients using third party references to catastrophise their seizure experiences are more likely to have PNES, whilst patients who use third party references to normalise their life with seizures are more likely to have epilepsy.