Intonation Units in Spontaneous Speech Evoke a Neural Response.

Spontaneous speech is produced in chunks called intonation units (IUs). IUs are defined by a set of prosodic cues and presumably occur in all human languages. Recent work has shown that across different grammatical and sociocultural conditions IUs form rhythms of ∼1 unit per second. Linguistic theory suggests that IUs pace the flow of information in the discourse. As a result, IUs provide a promising and hitherto unexplored theoretical framework for studying the neural mechanisms of communication. In this article, we identify a neural response unique to the boundary defined by the IU. We measured the EEG of human participants (of either sex), who listened to different speakers recounting an emotional life event. We analyzed the speech stimuli linguistically and modeled the EEG response at word offset using a GLM approach. We find that the EEG response to IU-final words differs from the response to IU-nonfinal words even when equating acoustic boundary strength. Finally, we relate our findings to the body of research on rhythmic brain mechanisms in speech processing. We study the unique contribution of IUs and acoustic boundary strength in predicting delta-band EEG. This analysis suggests that IU-related neural activity, which is tightly linked to the classic Closure Positive Shift (CPS), could be a time-locked component that captures the previously characterized delta-band neural speech tracking.SIGNIFICANCE STATEMENT Linguistic communication is central to human experience, and its neural underpinnings are a topic of much research in recent years. Neuroscientific research has benefited from studying human behavior in naturalistic settings, an endeavor that requires explicit models of complex behavior. Usage-based linguistic theory suggests that spoken language is prosodically structured in intonation units. We reveal that the neural system is attuned to intonation units by explicitly modeling their impact on the EEG response beyond mere acoustics. To our understanding, this is the first time this is demonstrated in spontaneous speech under naturalistic conditions and under a theoretical framework that connects the prosodic chunking of speech, on the one hand, with the flow of information during communication, on the other.

Changes in the structure of spontaneous speech predict the disruption of hierarchical brain organization in first-episode psychosis.

Psychosis implicates changes across a broad range of cognitive functions. These functions are cortically organized in the form of a hierarchy ranging from primary sensorimotor (unimodal) to higher-order association cortices, which involve functions such as language (transmodal). Language has long been documented as undergoing structural changes in psychosis. We hypothesized that these changes as revealed in spontaneous speech patterns may act as readouts of alterations in the configuration of this unimodal-to-transmodal axis of cortical organization in psychosis. Results from 29 patients with first-episodic psychosis (FEP) and 29 controls scanned with 7 T resting-state fMRI confirmed a compression of the cortical hierarchy in FEP, which affected metrics of the hierarchical distance between the sensorimotor and default mode networks, and of the hierarchical organization within the semantic network. These organizational changes were predicted by graphs representing semantic and syntactic associations between meaningful units in speech produced during picture descriptions. These findings unite psychosis, language, and the cortical hierarchy in a single conceptual scheme, which helps to situate language within the neurocognition of psychosis and opens the clinical prospect for mental dysfunction to become computationally measurable in spontaneous speech.

'Should we laugh?' Acoustic features of (in)voluntary laughters in spontaneous conversations.

Laughter is one of the most common non-verbal features; however, contrary to the previous assumptions, it may also act as signals of bonding, affection, emotional regulation agreement or empathy (Scott et al. Trends Cogn Sci 18:618-620, 2014). Although previous research agrees that laughter does not form a uniform group in many respects, different types of laughter have been defined differently by individual research. Due to the various definitions of laughter, as well as their different methodologies, the results of the previous examinations were often contradictory. The analysed laughs were often recorded in controlled, artificial situations; however, less is known about laughs from social conversations. Thus, the aim of the present study is to examine the acoustic realisation, as well as the automatic classification of laughter that appear in human interactions according to whether listeners consider them to be voluntary or involuntary. The study consists of three parts using a multi-method approach. Firstly, in the perception task, participants had to decide whether the given laughter seemed to be rather involuntary or voluntary. In the second part of the experiment, those sound samples of laughter were analysed that were considered to be voluntary or involuntary by at least 66.6% of listeners. In the third part, all the sound samples were grouped into the two categories by an automatic classifier. The results showed that listeners were able to distinguish laughter extracted from spontaneous conversation into two different types, as well as the distinction was possible on the basis of the automatic classification. In addition, there were significant differences in acoustic parameters between the two groups of laughter. The results of the research showed that, although the distinction between voluntary and involuntary laughter categories appears based on the analysis of everyday, spontaneous conversations in terms of the perception and acoustic features, there is often an overlap in the acoustic features of voluntary and involuntary laughter. The results will enrich our previous knowledge of laughter and help to describe and explore the diversity of non-verbal vocalisations.

The characteristics of spontaneous language in young children identified as language delayed in Mandarin.

BACKGROUND: Little is known about the spontaneous speech characteristics of young children with language delay in Mandarin, relative to their peers. Until the recent development of standardized language assessments normed in China on Mandarin-speaking children, it was difficult to independently identify atypically developing children to study their spontaneous speech, and only case studies have been available. AIM: To investigate which aspects of spontaneous speech might be distinctive for atypical development in a short play session. METHODS & PROCEDURES: A total of 86 Mandarin-speaking children, boys and girls aged 2;6-4;6, were tested using the new standardized assessment for Mandarin, Diagnostic Receptive Expressive Assessment of Mandarin-Comprehensive (DREAM-C), at a major urban hospital in China. Of the children, 39 were identified by DREAM-C as atypically developing in language development (Total Standard score M = 72, SD = 8.9), and 47 scored in the typical range (Total Standard score M = 103, SD = 10.8), using the four scales of Receptive, Expressive Semantics and Syntax. All children then took part in a 15-min semi-structured play session during which their spontaneous speech was recorded by professionals. A variety of games and pictures were used in an attempt to elicit spontaneous questions, negatives and descriptions. Their recorded speech samples were then coded by linguists directly into a database in FilemakerPro for different aspects of vocabulary, sentence variety and grammatical morphemes/structures heard. OUTCOMES & RESULTS: The results describe the characteristics of the speech samples for the typical and atypical groups for age groups 2;6-3.5 (N = 52) and 3;6-4.6 (N = 34). Vocabulary diversity was indicated on an ordinal scale ranging from simple communicative signals including headshakes and words such as 'hi' to 'a rich variety of different content words'. Grammatical diversity similarly ranged from 'only yes/no answers', through to the appearance of multi-clause sentences. Morphosyntax was coded in terms of which morphemes were observed at all in the session, such as aspect markers (LE, ZAI, GUO), and nominal morphemes (DE, GE), and also whether function words such as pronouns, Wh-questions and classifiers were singular or varied in the session. There is considerable optionality in morpheme expression in Mandarin, so measurements such as the percentage supplied in obligatory contexts, though useful for languages such as English, are harder to compute. Nevertheless, the data show change over age in all these aspects of language, and reveal what a typically developing child might be expected to produce in a 15-min sample in such a session. For example, it was rare for the typically developing children by age 3.5 not to have at least simple sentences with some function words, and to use adjectives, nouns and verbs, unlike the children with atypical scores. The morphemes DE, LE and GUO showed significant differences in likelihood across groups for both ages, but BA and ZAI were significantly different only for the older age group. In contrast, GE was common in all groups. The atypical group has markedly lower frequency in several grammatical aspects such as the use of diverse questions, classifiers and pronouns, with much less change across age groups, implying slower growth. The results provide useful information on the relative likelihood of observing different varieties of words, sentence types and morphemes in a short speech sample, which are substantially different in the typical versus atypical groups in both age bands, and change over age. CONCLUSIONS & IMPLICATIONS: These data validate the DREAM-C classifications, but the details can also be used to inform the choice of targets for intervention for young children who experience delays in Mandarin language acquisition. WHAT THIS PAPER ADDS: What is already known on the subject Spontaneous language samples have been used as a means of studying language in China. However, because of the amount of training and time required to transcribe and analyse spontaneous language samples, there is not yet a sufficient basis for identifying language differences between children with and without language disorder in mainland China. What does this paper add to existing knowledge After using DREAM-C to provide an objective measurement of children with and without language disorders, an easy-to-administer spontaneous language assessment protocol and scoring record form allowed the comparison of the spontaneous language of 39 children with atypically developing language with 47 children who scored in the typical range to observe language differences between those with and without typical language development. What are the potential or actual clinical implications of this work? The study documents differences in vocabulary and sentence variety, and use of different morphemes such as classifiers, passives and aspect markers to aid in choosing targets for intervention by demonstrating the path of development. In addition, the spontaneous language assessment protocol and scoring record form holds promise for allowing clinicians and researchers to more easily study the language of individual children to personalize intervention, but also of groups of children to understand the emergence of basic Mandarin linguistic features.

Spontaneous speech intelligibility: early cochlear implanted children versus their normally hearing peers at seven years of age.

Speaking intelligibly is an important achievement in children's language development. How far do congenitally severe-to-profound hearing-impaired children who received a cochlear implant (CI) in the first two years of their life advance on the path to intelligibility in comparison to children with typical hearing (NH)? Spontaneous speech samples of children with CI and children with NH were orthographically transcribed by naïve transcribers. The entropy of the transcriptions was computed to analyze their degree of uniformity. The same samples were also rated on a continuous rating scale by another group of adult listeners. The transcriptions of the NH children's speech were more uniform, i.e., had significantly lower entropy, than those of the CI children, suggesting that the latter group displayed lower intelligibility. This was confirmed by the ratings on the continuous scale. Despite the relatively restricted age ranges, older children reached better intelligibility scores in both groups.

Developmental language disorder in sequential bilinguals: Characterising word properties in spontaneous speech.

The current study sought to investigate whether word properties can facilitate the identification of developmental language disorder (DLD) in sequential bilinguals by analyzing properties in nouns and verbs in L2 spontaneous speech as potential DLD markers. Measures of semantic (imageability, concreteness), lexical (frequency, age of acquisition) and phonological (phonological neighbourhood, word length) properties were computed for nouns and verbs produced by 15 sequential bilinguals (5;7) with DLD and 15 age-matched controls with diverse L1 backgrounds. Linear mixed modelling revealed a significant interaction of group and word category on phonological neighbourhood values but no differences across imageability, concreteness, frequency, age of acquisition, and word length measures in spontaneous speech. Outcomes suggest that group-level differences may not be apparent at the word-level, due to the heterogeneous nature of DLD and potential similarities in production during early L2 acquisition.

E-Prevention: Advanced Support System for Monitoring and Relapse Prevention in Patients with Psychotic Disorders Analyzing Long-Term Multimodal Data from Wearables and Video Captures.

Wearable technologies and digital phenotyping foster unique opportunities for designing novel intelligent electronic services that can address various well-being issues in patients with mental disorders (i.e., schizophrenia and bipolar disorder), thus having the potential to revolutionize psychiatry and its clinical practice. In this paper, we present e-Prevention, an innovative integrated system for medical support that facilitates effective monitoring and relapse prevention in patients with mental disorders. The technologies offered through e-Prevention include: (i) long-term continuous recording of biometric and behavioral indices through a smartwatch; (ii) video recordings of patients while being interviewed by a clinician, using a tablet; (iii) automatic and systematic storage of these data in a dedicated Cloud server and; (iv) the ability of relapse detection and prediction. This paper focuses on the description of the e-Prevention system and the methodologies developed for the identification of feature representations that correlate with and can predict psychopathology and relapses in patients with mental disorders. Specifically, we tackle the problem of relapse detection and prediction using Machine and Deep Learning techniques on all collected data. The results are promising, indicating that such predictions could be made and leading eventually to the prediction of psychopathology and the prevention of relapses.

Fluency in L2: Read and Spontaneous Speech Pausing Patterns of Turkish, Swahili, Hausa and Arabic Speakers of English.

Language learners' actual speech performances constitute an essential aspect of studies on second language learning and teaching. Although there is ample research on fluency and pauses in English, current literature does not touch on this issue from a multilingual perspective by comparing both read and spontaneous speech performances. In this descriptive study, the researchers investigated pausing patterns with 40 Turkish, Swahili, Hausa, and Arabic speakers of English. For the read speech fragments' elicitation, the participants read out a short story, and for spontaneous speech, the data was gathered through structured interviews. In total, 4007 pauses were measured through Praat, and the findings obtained from the data were analyzed using multiple regression and several multivariate analyses of variance. The findings revealed crucial insights into the nature of fluency research in terms of (a) speech registers, (b) positions, (c) conjunctions, and (d) mother tongues.

Long-term effects of cochlear implantation on the intelligibility of speech in French-speaking children.

Our study compares the intelligibility of French-speaking children with a cochlear implant (N = 13) and age-matched children with typical hearing (N = 13) in a narrative task. This contrasts with previous studies in which speech intelligibility of children with cochlear implants is most often tested using repetition or reading tasks. Languages other than English are seldom considered. Their productions were graded by naive and expert listeners. The results show that (1) children with CIs have lower intelligibility, (2) early implantation is a predictor of good intelligibility, and (3) late implantation after two years of age does not prevent the children from eventually reaching good intelligibility.

Changes in temporal features of speech in secondary progressive multiple sclerosis (SPMS) - case studies.

Multiple sclerosis (MS) is a chronic inflammatory disease which, in addition to affecting motor and cognitive functions, may involve language disorders. Despite the importance of speech and language disorders in the quality of life of patients, there are only a few studies about language and speech production difficulties in MS. The aim of this research is to describe the limitation patterns of speech and temporal characteristics of the suprasegmental level in two SPMS cases related to various types of spontaneous speech tasks. We assumed the change of the cognitive load has a greater effect on spontaneous speech in MS patients than in controls. Two SPMS patients, and two sex-, age- and education matched healthy controls were studied. We applied verbal fluency tests (phonemic, episodic, semantic, verb), digit span test, non-word repetition test, Corsi Block Tapping Test, Stroop Colour and Word Test, and Trail Making Test. Token Test was used to measure speech comprehension. The four speech tasks required relatively different degrees of cognitive effort: (a) spontaneous narrative about own life; (b) event description; (c) picture description; (d) narrative recall. Our results show that there are differences between MS patients and controls: MS patients produced slower speech and articulation rate, and they had more and longer pauses in every speech task. Speech tasks and the degree of the cognitive load had a greater effect on MS patients than on control speakers.

Automated screening for distress: A perspective for the future.

Distress is a complex condition, which affects a significant percentage of cancer patients and may lead to depression, anxiety, sadness, suicide and other forms of psychological morbidity. Compelling evidence supports screening for distress as a means of facilitating early intervention and subsequent improvements in psychological well-being and overall quality of life. Nevertheless, despite the existence of evidence-based and easily administered screening tools, for example, the Distress Thermometer, routine screening for distress is yet to achieve widespread implementation. Efforts are intensifying to utilise innovative, cost-effective methods now available through emerging technologies in the informatics and computational arenas.

Early Recovery of Aphasia through Thrombolysis: The Significance of Spontaneous Speech.

BACKGROUND: Aphasia is one of the most devastating stroke-related consequences for social interaction and daily activities. Aphasia recovery in acute stroke depends on the degree of reperfusion after thrombolysis or thrombectomy. As aphasia assessment tests are often time-consuming for patients with acute stroke, physicians have been developing rapid and simple tests. The aim of our study is to evaluate the improvement of language functions in the earliest stage in patients treated with thrombolysis and in nontreated patients using our rapid screening test. MATERIALS AND METHODS: Our study is a single-center prospective observational study conducted at the Stroke Unit of the University Medical Hospital of Trieste (January-December 2016). Patients treated with thrombolysis and nontreated patients underwent 3 aphasia assessments through our rapid screening test (at baseline, 24 hours, and 72 hours). The screening test assesses spontaneous speech, oral comprehension of words, reading aloud and comprehension of written words, oral comprehension of sentences, naming, repetition of words and a sentence, and writing words. RESULTS: The study included 40 patients: 18 patients treated with thrombolysis and 22 nontreated patients. Both groups improved over time. Among all language parameters, spontaneous speech was statistically significant between 24 and 72 hours (P value = .012), and between baseline and 72 hours (P value = .017). CONCLUSIONS: Our study demonstrates that patients treated with thrombolysis experience greater improvement in language than the nontreated patients. The difference between the 2 groups is increasingly evident over time. Moreover, spontaneous speech is the parameter marked by the greatest improvement.

Informativeness, Timing and Tempo in Lexical Self-Repair.

This paper presents a study of the temporal organization of lexical repair in spontaneous Dutch speech. It assesses the extent to which offset-to-repair duration and repair tempo can be predicted on the basis of offset timing, reparandum tempo and measures of the informativeness of the crucial lexical items in the repair. Specifically, we address the expectations that repairs that are initiated relatively early are produced relatively fast throughout, and that relatively highly informative repairs are produced relatively slowly. For informativeness, we implement measures based on repair semantics, lexical frequency counts and cloze probabilities. Our results highlight differences between factual and linguistic error repairs, which have not been consistently distinguished in previous studies, and provide some evidence to support the notion that repairs that are initiated relatively early are produced relatively fast. They confirm that lexical frequency counts are rough measures of contextual predictability at best, and reveal very few significant effects of our informativeness measures on the temporal organization of lexical self-repair. Moreover, although we can confirm that most repairs have a repair portion that is fast relative to its reparandum, this cannot be attributed to the relative informativeness of the two portions. Our findings inform the current debate on the division of labour between inner and overt speech monitoring, and suggest that, although the influence of informativeness on speech production is extensive, it is not ubiquitous.

Dynamic aphasia following low-grade glioma surgery near the supplementary motor area: a selective spontaneous speech deficit.

We describe a patient (KO) with reduced spontaneous speech, resembling dynamic aphasia, after awake glioma surgery in the proximity of the supplementary motor area. Naming, repetition, and comprehension were intact. He was tested with an extensive neuropsychological test-battery and a protocol for dynamic aphasia at 1 year. He presented with postoperative reduced spontaneous speech and selective executive function deficits. Most language recovery took place at 3 months postoperatively, whereas the executive functions improved between 3 months and 1 year. Results suggest that resection near the supplementary motor area could increase the risk of cognitive disturbances at long term, especially language.

Assessment of intelligibility using children's spontaneous speech: methodological aspects.

BACKGROUND: Intelligibility is a speaker's ability to convey a message to a listener. Including an assessment of intelligibility is essential in both research and clinical work relating to individuals with communication disorders due to speech impairment. Assessment of the intelligibility of spontaneous speech can be used as an overall indicator of the severity of a speech disorder. There is a lack of methods for measuring intelligibility on the basis of spontaneous speech. AIMS: To investigate the validity and reliability of a method where listeners transcribe understandable words and an intelligibility score is calculated on the basis of the percentage of syllables perceived as understood. METHODS & PROCEDURES: Spontaneous speech from ten children with speech-sound disorders (mean age = 6.0 years) and ten children with typical speech and language development (mean age = 5.9 years) was recorded and presented to 20 listeners. Results were compared between the two groups and correlation with percentage of consonants correct (PCC) was examined. OUTCOMES & RESULTS: The intelligibility scores obtained correlated with PCC in single words and differed significantly between the two groups, indicating high validity. Inter-judge reliability, analysed using intra-class correlation (ICC), was excellent in terms of the average measure for several listeners. CONCLUSIONS & IMPLICATIONS: The results suggest that this method can be recommended for assessing intelligibility, especially if the mean across several listeners is used. It could also be used in clinical settings when evaluating intelligibility over time, provided that the same listener makes the assessments.

Neural activation during natural speech and rests in patients with schizophrenia and schizophrenia spectrum disorders-an fMRI pilot trial.

Background: In schizophrenia patients, spontaneous speech production has been hypothesized as correlating with right hemispheric activation, including the inferior frontal and superior temporal gyri as speech-relevant areas. However, robust evidence for this association is still missing. The aim of the present fMRI study is to examine BOLD signal changes during natural, fluent speech production in patients with schizophrenia in the chronic phase of their disease. Methods: Using a case-control design, the study included 15 right-handed patients with schizophrenia spectrum disorders as well as 15 healthy controls. The participants described eight pictures from the Thematic Apperception Test for 1 min each, while BOLD signal changes were measured with 3T fMRI. The occurrence of positive and negative formal thought disorders was determined using standardized psychopathological assessments. Results: We found significant BOLD signal changes during spontaneous speech production in schizophrenia patients compared to healthy controls, particularly in the right hemispheric network. A post-hoc analysis showed that this right-hemispheric lateralization was mainly driven by activation during experimental rests. Furthermore, the TLI sum value in patients correlated negatively with BOLD signal changes in the right Rolandic operculum. Conclusions: Possible underlying factors for this inverse right-hemispheric lateralization of speech-associated areas are structural changes and transmitter system alterations, as well as a lack of neural downregulation in schizophrenia patients during rest periods due to dysfunctional executive functions. When examining spontaneous speech as the most natural form of language, other influencing factors, such as social cognition or emotional processing, should be considered. Our results indicate that future studies should consider group differences during rest, which might provide additional information typically covered in differential contrasts.

Young Children's Directed Question Asking in Preschool Classrooms.

Question asking is a prevalent aspect of children's speech, providing a means by which young learners can rapidly gain information about the world. Previous research has demonstrated that children exhibit sensitivity to the knowledge state of potential informants in laboratory settings. However, it remains unclear whether and how young children are inclined to direct questions that support learning deeper content to more knowledgeable informants in naturalistic classroom contexts. In this study, we examined children's question-asking targets (adults, other preschoolers, self-talk) during an open-play period in a US preschool classroom and assessed how the cognitive and linguistic characteristics of questions varied as a function of the intended recipient. Further, we examined how these patterns changed with age. We recorded the spontaneous speech of individual children between the ages of 3 and 6 years (N = 30, totaling 2875 utterances) in 40-min open-period sessions in their preschool day, noting whether the speech was directed toward an adult, another child, or was stated to self. We publish this fully transcribed database with contextual and linguistic details coded as open access to all future researchers. We found that questions accounted for a greater proportion of preschoolers' adult-directed speech than of their child-directed and self-directed speech, with a particular increase in questions that supported broader learning goals when directed to an adult. Younger children directed a higher proportion of learning questions to adults than themselves, whereas older children asked similar proportions of questions to both, suggesting a difference in younger and older children's question-asking strategies. Although children used greater lexical diversity in questions than in other utterances, their question formulation in terms of length and diversity remained consistent across age and recipient types, reflecting their general linguistic abilities. Our findings reveal that children discriminately choose "what" and "whom" to ask in daily spontaneous conversations. Even in less-structured school contexts, preschoolers direct questions to the informant most likely to be able to provide an adequate answer.

Adynamic spoken language in corpus callosum dysgenesis.

Corpus callosum dysgenesis (CCD) is a congenital brain malformation that occurs when the development of the corpus callosum is disrupted, either partially or completely. The cognitive outcomes in individuals with CCD vary greatly, but generally the neuropsychological profile is characterised by slow processing speed, poor transfer of interhemispheric sensory-motor information, and impaired complex problem solving. Core language skills are often preserved in CCD, but there is some evidence that complex language may be impaired. Thus, the current study sought to examine whether spontaneous speech output was reduced in a cohort of individuals with CCD compared to age-matched controls. We further explored a series of factors that may be contributing to poor spontaneous speech in CCD, such as difficulties generating, selecting, and sequencing ideas for expression, as well as apathy and slowed processing speed. A cohort of 25 individuals with CCD and 39 neurotypical controls were enrolled in this study. Participants completed a picture description task to measure spontaneous speech output, alongside a series of cognitive and language baseline tests. Verbal and nonverbal fluency tasks gauged idea generation and sequencing, and sentence-level selection tasks measured idea selection. We found that, despite having largely intact core language skills, individuals with CCD produced significantly less spontaneous speech on the picture description task than controls. This language profile may be described as "adynamic". Further, we found that poor spontaneous speech output in CCD was related to problems generating ideas for expression, as individuals with CCD performed below controls on the verbal and nonverbal fluency tasks. Exploratory analyses revealed that apathy and slowed processing speed may be contributing factors. Adynamia in CCD is a novel finding that may be an intervention target for improving communication skills in this population.

Distinguishable features of spontaneous speech in Alzheimer's clinical syndrome and healthy controls.

There is growing evidence that subtle changes in spontaneous speech may reflect early pathological changes in cognitive function. Recent work has found that lexical-semantic features of spontaneous speech predict cognitive dysfunction in individuals with mild cognitive impairment (MCI). The current study assessed whether Ostrand and Gunstad's (OG) lexical-semantic features extend to predicting cognitive status in a sample of individuals with Alzheimer's clinical syndrome (ACS) and healthy controls. Four additional (New) speech indices shown to be important in language processing research were also explored in this sample to extend prior work. Speech transcripts of the Cookie Theft Task from 81 individuals with ACS (Mage = 72.7 years, SD = 8.80, 70.4% female) and 61 healthy controls (HC) (Mage = 63.9 years, SD = 8.52, 62.3% female) from Dementia Bank were analyzed. Random forest and logistic machine learning techniques examined whether subject-level lexical-semantic features could be used to accurately discriminate those with ACS from HC. Results showed that logistic models with the New lexical-semantic features obtained good classification accuracy (78.4%), but the OG features had wider success across machine learning model types. In terms of sensitivity and specificity, the random forest model trained on the OG features was the most balanced. Findings from the current study suggest that features of spontaneous speech used to predict MCI may also distinguish between individuals with ACS and healthy controls. Future work should evaluate these lexical-semantic features in pre-clinical persons to further explore their potential to assist with early detection through speech analysis.

Machine Learning-Based Prediction Models for Cognitive Decline Progression: A Comparative Study in Multilingual Settings Using Speech Analysis.

Background: Mild cognitive impairment (MCI) is a condition commonly associated with dementia. Therefore, early prediction of progression from MCI to dementia is essential for preventing or alleviating cognitive decline. Given that dementia affects cognitive functions like language and speech, detecting disease progression through speech analysis can provide a cost-effective solution for patients and caregivers. Design-Participants: In our study, we examined spontaneous speech (SS) and written Mini Mental Status Examination (MMSE) scores from a 60-patient dataset obtained from the Mugla University Dementia Outpatient Clinic (MUDC) and a 153-patient dataset from the Alzheimer's Dementia Recognition through Spontaneous Speech (ADRess) challenge. Our study, for the first time, analyzed the impact of audio features extracted from SS in distinguishing between different degrees of cognitive impairment using both an Indo-European language and a Turkic language, which exhibit distinct word order, agglutination, noun cases, and grammatical markers. Results: When each machine learning model was tested on its respective trained language, we attained a 95% accuracy using the random forest classifier on the ADRess dataset and a 94% accuracy on the MUDC dataset employing the multilayer perceptron (MLP) neural network algorithm. In our second experiment, we evaluated the effectiveness of each language-specific machine learning model on the dataset of the other language. We achieved accuracies of 72% for English and 76% for Turkish, respectively. Conclusion: These findings underscore the cross-language potential of audio features for automated tracking of cognitive impairment progression in MCI patients, offering a convenient and cost-effective option for clinicians or patients.