Growing evidence demonstrates that subtle changes in spontaneous speech can be used to distinguish older adults with and without cognitive impairment, including those with Alzheimer's disease (AD). Recent work suggests that quantification of the meaningful connectedness of speech - termed semantic coherence - may be sensitive to cognitive dysfunction. The current study compared global coherence (GC; the degree to which individual utterances relate to the overall topic being discussed) and local coherence (LC; the degree to which adjoining utterances relate meaningfully to each other) in persons with AD and healthy controls.

Speech transcripts from 81 individuals with probable AD (Mage = 72.7 years, SD = 8.8, 70.3% female) and 61 healthy controls (HC) (Mage = 63.9 years, SD = 8.5, 62.2% female) from Dementia Bank were analyzed. All participants completed the Cookie Theft and MMSE as part of that larger project. Machine learning analyses of GC and LC were conducted and models evaluated classification accuracy (i.e., AD vs HC) as well as ROC-AUC. Relationships between coherence indices and MMSE performance were also quantified.

Though no significant group differences emerged in LC (Estimate = 0.012, p = 0.32), persons with AD differed from healthy controls in GC (Estimate = 0.03, p = 0.006) and produced less semantically coherent speech. GC indices predicted AD diagnoses with 65% accuracy. Interestingly, coherence indices showed only modest correlation with MMSE scores (r = .19).

GC metrics of spontaneous speech differentiated between persons with AD and controls, but did not strongly correlate with MMSE performance. Such findings support the notion that many aspects of language are impacted in persons with AD. In addition to replication, future work should evaluate whether GC is also disrupted in persons with pre-clinical AD and its potential to assist with early detection.

The Children’s Communication Checklist-Second Edition (CCC-2) is a rating scale designed to assess domains of communication skills with emphasis on pragmatics (Bishop, 2006). The CCC comprises 10 subtests addressing various aspects of oral communication skills: Speech, Syntax, Semantics, Coherence, Initiation, Scripted Language, Context, Nonverbal Communication, Social Relations, and Interests. In a study conducted on the original CCC, Geurts et al. (2004) found that when compared to normal controls, pragmatic difficulties occurred in children with either high functioning autism (HFA) or ADHD. Since the initial version of the CCC, no study has examined whether the revised version can differentiate children with HFA, ADHD, and LD, the purpose of the present study. Focus was on derived factors of the structure/content of language and the pragmatics of language.

Forty-one participants grouped according to diagnosis were drawn from two archival data pools, one adapted from a previous study conducted by Casey and Scott (2016) and the other from a set of anonymized patients from a neuropsychological clinic. Fourteen participants met clinical criteria for autism (Mage = 11.95), 12 participants met criteria for ADHD without co-morbid disorders (Mage = 9.5), and 15 participants met criteria for a learning disability involving reading, writing, math, or some combination (Mage = 10.13). Group-specific descriptive statistics were computed for the participants’ age, full scale intelligence quotient (IQ), and General Communication Composite (GCC). Two factor scores were computed, one composed of the subtests that constitute the structure/content aspects of language (Speech, Syntax, Semantics, and Coherence) and one composed of the pragmatic aspects of language (Initiation, Nonverbal Communication, Social Relations, and Interests), an area of particular weakness in HFA. Independent samples ANOVAs were conducted on both factor scores to determine whether the CCC-2 could differentiate the three groups. Post-hoc comparisons were planned for the subtests comprising the factor(s) that differentiated the groups.

Participants in the ADHD (M = 9.45, SD = 2.45) group were significantly younger than those in the HFA group (M = 11.95, SD = 2.24) and LD group (M = 10.13, SD = 2.58), the latter two not differing significantly. The groups did not differ significantly on IQ, nor on the structure/content factor. On the pragmatic factor, the LD group (M = 10.18, SD = 9.91) had significantly higher scores than the ADHD group (M = 7.79, SD = 6.54), which in turn, had significantly higher scores than the HFA group (M = 5.48, SD = 8.26), F(2, 38) = 17.81, p < .01. Within this composite, the same pattern was shown on Nonverbal Communication, F(2, 38) = 9.29, p < .01, and Interests, F(2, 38) = 17.81, p < .01.

Compared to children with an academically-based learning disability, children with ADHD and HFA demonstrated pragmatic difficulties on the CCC-2. Although there was overlap between the pragmatic language characteristics of children with ADHD and children with HFA, the CCC-2 demonstrated utility in distinguishing the two disorders on certain aspects of communication skills, suggesting that it is a useful tool in differential diagnosis.

Brain science demonstrates that people who stutter (PWS) exhibit insufficient activation in the auditory speech area of the left hemisphere (Kikuchi, et al. 2011 ; Garnett, et al. 2018). In this study, we reported the auditory brainstem response of PWS: in PWS with moderate and severe impairment, significantly longer interpeak latencies (IPLs) between waves I and V (IPL [I-V]) of the right ear than those of the left ear were observed. However, in PWS with mild impairment, the IPLs (I-V) of the left ear were significantly longer than those of the right ear (Anzaki et al., 2020). We considered that the differences in the IPLs (I-V) between the right and left ears cause monitoring disturbance in communication, which results in developmental stuttering. It has been reported that stuttering was improved by delayed auditory feedback (DAF) (Stromsta, 1956; Sakai, 2008). Thus, we improved the DAF system and developed an application that can be used by PWS to listen to their own voices with no differences in the IPLs (I-V) between their left and right ears. We verified the effectiveness of this application.

This study included five male adults with developmental stuttering (ADSs), with a mean age and handedness index of 36 years and 84, respectively. The application was adjusted so that the IPLs (I-V) of the left and right ears were the same. For example, one ADS showed that the IPL (I-V) of their right ear was 0.5 msec longer than that of their left. Subsequently, the application was adjusted so that the IPL (I-V) of his left ear would be delayed by 0.5 msec. We asked the participants to use the application for six months when free talking and reading aloud. Using the Japanese Standardized Test for Stuttering (JSTS) (Ozawa, et al. 2013), we compared their disfluencies with and without the application.

As per the JSTS, the stuttering severity in all participants improved. Case 1, who had severe impairment (level 5), showed a moderate improvement (level 4), Cases 2 and 3, who had moderate impairment (level 4), showed a mild improvement (level 3), and Cases 4 and 5, who had mild impairment (level 3), exhibited a normal level of improvement (level 1). We calculated the z-scores of the improvement rates of the JSTS based on the standard deviations according to the severity (Anzaki, 2019). The z-scores of Case 4 and 5 were 4.01 and 2.01, respectively, indicating a significant improvement.

In our report last year, although ADSs with moderate and severe impairment showed improvement by stimulation intervention on the left hemisphere through the right ear, those with mild impairment exhibited only a slight or no improvement as per the JSTS (Anzaki, et al. 2021). The application developed in this study was found to significantly improve the disfluencies of all the participants as per the JSTS, especially those with mild impairment. Therefore, we considered that stuttering disorders are layered; ADSs have auditory monitoring disorder in the base.

Early detection is critical to the effective management of Alzheimer’s Disease (AD) and other forms of dementia. Existing screening assessments are often costly, require substantial expertise to administer, and may be insensitive to mild changes in cognition. A promising alternative is automatically measuring features of connected speech (c.f., Ostrand & Gunstad, 2021, Journal Geriatric Psych & Neurol) to predict impairment. Here, we built on prior work examining how well speech features predicted cognitive impairment. Unique to the current work, we attempted to capture more holistic effects of cognitive impairment by examining the relevance of linguistic features that measure sentential or discourse context properties of speech, including the context in which filler words (e.g., um) occur, and the predictability of individual words within their sentence context, computed from a large computational language model (GPT-2).

Participants completed the Cookie Theft picture description task, with data available in the DementiaBank corpus (Becker et al., 1994, Arch Neurol). Descriptions that contained at least 50 words (N = 214) were submitted to an automatic feature calculation pipeline written in Python to calculate various part-of-speech counts, lexical diversity metrics, and mean lexical frequency, as well as multiple metrics related to lexical surprisal (i.e., how surprising a word is given its context).

Surprisal of individual words was computed using the pre-trained GPT-2 transformer language model (Radford et al., 2019, Comput. Sci.) by computing word probability given the previous 12 words. Multiple linear regression was performed using 17 linguistic features jointly as predictors, and Mini-Mental State Examination (MMSE) score as the outcome variable. Simple regressions were calculated between each feature and MMSE scores to examine the predictability of each linguistic feature on cognitive decline.

A multiple linear regression model containing all linguistic features plus demographic information (age, sex, education) significantly predicted MMSE scores (Adjusted R2 = 0.41, F20, 193 = 8.37, p < .001), and explained significantly more variance in MMSE scores than did demographic variables alone (F17193 = 6.85, p < .001). Individual predictors that were significantly correlated with MMSE score included: how unexpected an individual’s word choices were, given the preceding context (median surprisal: r = -0.33, p < 0.001; interquartile range: r = 0.18, p < 0.02), mean lexical frequency (r = -0.50, p < .001), and usage of definite articles (r = 0.31, p < 0.001), nouns (r = 0.26, p < .001), and empty words (r = -0.25, p < 0.001).

Participants with lower MMSE scores, indicating greater impairment, used more frequent, yet more surprising, words, and produced more empty words and fewer definite articles and nouns. These results suggest that measures of semantic specificity and coherence of speech could be meaningful predictors of cognitive decline, and can be computed automatically from speech transcriptions. The results also provide novel evidence that computational approaches to estimating lexical predictability may have value in predicting the degree of decline, motivating future work in other speech elicitation tasks and differing clinical groups.

Schizophrenia is often disguised as a crisis of adolescence. We want to understand how schizophrenia manifests itself in speech. We expect the difference in the Word Fluency Association Task (WFAT) in Normality Index and Time Delay.

The analyses data was collected by the WFAT (authored by V.Kritskaya). WFAT actualize speech connections based on past experience. Stimuli - various syllables of 2-letters (20 pcs.), 3-letters (30 pcs.), varying in frequency of use (compiled on the basis of the Corpuses of the Russian language). Instruction: “Now I will tell you a syllable, your task is to complete the proposed syllable to words as quickly as possible, your words must be real”. For study purposes we subdivided the sample into two subgroups: 12-14 years, 15-17 years. For statistical analysis we used U-criterion, Mann-Whitney. Analyzed parameters:

Normality index (NI) - the ratio of productive nouns to the amount of the standard associations

Time delay (TD) - response delay of the subject (sec., millisec.)

Study involved 57 participants: 27 adolescents with schizophrenia (Cl_G) aged from 12 to 18 years, assessed in the hospital setting (DS: F20.xx, F21.xx) and 30 normotypal peers (Co_G) assessed in the school setting. Exclusion criteria: patients in the acute psychotic phase, left-handers.

NI grouped with the following rates: subgroup 1- Co_G 70,4, Cl_G 58,5, subgroup 2 - Co_G 65,2, Cl_G 56,1 There was no statistically significant difference between Cl_g Co_g in NI (P=0,000 and p=0,002 respectively). TD grouped with the following rates: subgroup 1-Co_G 2,42, Cl_G 22,63, subgroup 2 - Co_G 4,48, Cl_G 3,14 There was statistically significant difference between Cl_g Co_g in TD (P=0,432 and p=0,25 respectively).

Temporal indicators of cognitive activity differ in the clinical and control groups, which testifies to the significance of this indicator in the context of WFAT. There is no difference in terms of semantics. However, we expect to see it, because such difference has previously been shown in a group of adult patients with a severe type of schizophrenia. In the future, we would like to expand the group and select additional methods that evaluate the semantic component.