Duffy, Joseph R

Purpose: This study examined whether phoneme-level posterior probabilities generated by an articulatory-informed deep learning model (Phonet) can serve as interpretable and sensitive markers of articulatory impairment in motor speech disorders (MSDs). The goal was to determine whether these features distinguish individuals with and without MSDs and reflect varying severity levels of articulatory impairment. Method: A total of 676 participants (199 with MSDs, 477 without MSDs) completed a remotely captured word repetition task. Speech recordings were phonetically aligned, and posterior probabilities were extracted using a retrained version of the Phonet model, which estimates frame-level probabilities for 23 phonological classes. Segment-level posterior values were computed and compared across groups and severity levels using linear mixed-effects regression models. Results: Posterior probabilities were significantly reduced in the MSD group across all phoneme types. Within the MSD group, posterior probabilities decreased progressively with greater clinical severity, as measured by both perceptual sound distortion ratings and a cumulative articulation error score. Segment-specific analyses revealed systematic differences across phoneme classes, indicating that posterior probabilities capture distinct patterns of articulatory degradation associated with motor speech impairment. Conclusions: Phoneme-level posterior probabilities serve as interpretable, segment-specific indicators of articulatory impairment in MSDs. These features differentiate clinical status and reflect gradations of severity, making them promising tools for augmenting traditional perceptual speech assessments. Their linguistic transparency and scalability support future clinical applications in diagnosis, monitoring, and targeted intervention.

Abstract Purpose To evaluate whether two automated measures of nasality, Nasal Cognate Distinctiveness (NCD) and nasal posterior probability (NPP), provide clinically meaningful markers of hypernasality in speakers with motor speech disorders (MSDs) and to determine whether hypernasality-related information extends beyond stop consonants, which are typically targeted in cognate-based approaches. Method Sentence-level speech recordings were obtained from 561 patients, including 374 neurologically typical controls and 187 individuals with MSDs. NCD was computed for voiceless stops (/p/, /t/, /k/) using likelihood ratios between stops and their nasal cognates. NPP was extracted using a phonological feature-based neural network model and analyzed for stops, consonants, and vowels. Hypernasality severity was rated by speech-language pathologists. Group differences and associations with hypernasality severity were evaluated using linear mixed-effects models and Spearman correlations, and receiver operating characteristic (ROC) analyses were used to evaluate discrimination of absent versus present hypernasality. An additional 20 controls with repeat recordings were used to evaluate test-retest reliability and minimal detectable change (MDC). Results NCD generally decreased and NPP increased with increasing hypernasality severity, with phoneme-specific severity effects observed for NCD. Across stops, NCD and NPP were strongly inversely correlated and showed comparable discrimination of absent versus present hypernasality (AUCs = 0.796 and 0.730, respectively). Consonant and vowel NPP measures were both positively associated with hypernasality severity and demonstrated identical classification performance (AUC = 0.853). No significant differences in ROC performance were observed between NCD and NPP or between consonant and vowel NPP measures. Test-retest change exceeded the MDC threshold in 2 of 20 controls for NCD and 1 of 20 for NPP. Conclusions Both NCD and NPP provide objective markers of hypernasality in speakers with MSDs. The measures capture related aspects of nasalization and demonstrate comparable ability to identify clinically meaningful hypernasality. NPP captures hypernasality-related acoustic information across both consonants and vowels, with similar relationships to severity and classification performance across sound classes. Test-retest findings provide preliminary MDC benchmarks for interpreting longitudinal change.

Abstract Clinical syndromes of frontotemporal dementia present both distinct and overlapping neuroimaging abnormalities and, consequently, domains of impairment. The aim of this study was to determine how well covariance patterns from multiple neuroimaging modalities discriminate across clinical frontotemporal dementia syndromes. Four-hundred participants with a clinical frontotemporal dementia diagnosis, including behavioral variant of frontotemporal dementia, semantic variant of primary progressive aphasia, right temporal variant of frontotemporal dementia, primary progressive aphasia, apraxia of speech with agrammatic aphasia, primary progressive apraxia of speech, progressive supranuclear palsy and corticobasal syndrome, and 109 cognitively unimpaired participants underwent extensive neurological and neuropsychological assessments, structural MRI, flortaucipir-PET for tau and [18F] fluorodeoxyglucose-PET. Multimodal imaging covariance was investigated at the voxel-level using principal component analysis after adjusting for age and sex effects. Linear regression models were fit to investigate the relationship between principal components and clinical scores. Multinomial logistic regression models investigated the ability of imaging principal components to discriminate across syndromes. The first principal component, which describes the largest source of variability in the data, was a pattern of widespread cortical atrophy (21% variability explained) and of frontoparietal and temporal hypometabolism (21%): scores were lowest in behavioral variant of frontotemporal dementia and correlated with general cognitive impairment. On tau-PET, the first principal component (43% variability) was not significantly different among syndromes and captured a widespread cortical tau uptake, with a focus on the temporal lobe. In all three modalities, the second principal component captured temporal lobe imaging abnormalities, contrasting the motor and semantic clinical syndromes. Subsequent components described smaller variability. On cross-validation, the classifiers based on principal components from MRI (multiclass area under the receiver operating characteristic: 0.76) and [18F]fluorodeoxyglucose-PET (0.79) outperformed tau-PET (0.71), while combining all three modalities did not significantly improve the classifier performance. On cross-validation, the sensitivity of the image-based classifiers was heterogeneous across syndromes and modalities, with highest values for progressive supranuclear palsy and semantic variant of primary progressive aphasia and lowest for primary progressive aphasia and corticobasal syndrome. Our study found that the covariance components of structural and molecular neuroimaging underlying frontotemporal dementia clinical syndromes exist both as a continuum and with syndrome-specific patterns. More than one imaging model was able to correctly classify several frontotemporal dementia clinical syndromes with good specificity and sensitivity, highlighting the value of imaging in aiding patient diagnosis.

Abstract Motor speech disorders (MSDs) are early markers of neurological disease, but expert perceptual analysis is rarely available outside specialized centers. Automated speech analysis offers a scalable alternative, yet prior studies have not systematically compared modeling approaches or assessed clinically relevant metrics in independent datasets. This study compared static acoustic features, articulatory informed Phonet features, and self-supervised pretrained models for binary and multi label MSD classification. We trained and evaluated models on 583 speech samples using speaker level splits. Baseline models included logistic regression and Gated Recurrent Units (GRUs) trained on eGeMAPS and MFCCs. We extracted three types of Phonet derived features and evaluated pretrained HuBERT and SSAST models in frozen, partially fine-tuned, and fully fine-tuned configurations. Binary classification distinguished MSDs from controls, while multi label classification identified six MSD subtypes. Models were assessed using validation AUC, and cut points were tested on two independent datasets. Pretrained and Phonet based models substantially outperformed static acoustic features. In binary classification, HuBERT achieved the highest AUC (0.95), while compact Phonet derived GRUs achieved comparable performance (up to 0.94). These models generalized well to independent datasets, maintaining high sensitivity (0.94) and specificity (0.97). In multi label classification, Phonet models achieved the highest macro average AUC (0.86), but threshold-based subtype performance declined on unseen data. Automated MSD detection is feasible and clinically promising. Binary classification generalized well, whereas multi label classification showed limited threshold stability across datasets.

Abstract This study examined the extent to which goodness of pronunciation (GoP) scores and phonological posterior probabilities capture perceptual ratings of speech severity in individuals with motor speech disorders (MSD). Speech recordings of the word catastrophe were obtained from 489 participants, including 333 neurologically typical controls and 156 individuals with MSD. GoP scores were derived using traditional acoustic features and self-supervised speech representations, including WavLM and XLS-R, across multiple modeling approaches, while phonological posterior probabilities were extracted using Phonet. Model performance was evaluated using Kendall’s rank correlations, regression, and receiver operating characteristic analyses against speech-language pathologists’ perceptual ratings of sound distortion and intelligibility. Both GoP and phonological posterior probabilities were significantly associated with perceptual ratings. Self-supervised speech representations substantially outperformed traditional acoustic features, with WavLM-based GoP using k-nearest neighbors achieving the strongest performance. Across correlation, regression, and classification analyses, GoP consistently outperformed phonological posterior probabilities for both sound distortion and intelligibility. Age and gender had minimal influence on model-derived measures or their relationships with perceptual ratings. These findings demonstrate the value of self-supervised GoP as an objective measure of speech impairment while highlighting the complementary role of phonological posterior probabilities in characterizing articulatory aspects of motor speech disorders.

Purpose: This letter addresses a recent paper by Mailend et al. (2025) that expressed concerns about the Apraxia of Speech Rating Scale (ASRS) and offered guidelines and definitions regarding its principles, scale labels, item definitions, and scoring procedures, including the addition of new items to the scale. The primary purpose of this letter is to clarify the uncertainty that other users of the scale may have about whether the current ASRS version (ASRS 3.5; Duffy et al., 2023) should be altered to include the numerous proposed revisions. It provides examples of proposed revisions that, in our judgment, are helpful, redundant, questionable or premature, or are misaligned with the primary purposes of the scale. Conclusions: Some of the proposed revisions to the ASRS are helpful and likely to improve reliability by reducing user uncertainty about ratings. Others are questionable or premature without supporting data or are not aligned with the primary purposes of the scale.

Purpose: This viewpoint addresses the complexity of the nomenclature of progressive speech-language disorders, which often exists at the intersection of descriptive symptoms, clinical syndromes, and underlying neuropathology. Using primary progressive apraxia of speech (PPAOS) as an example, we explore how overlapping diagnostic frameworks employed by speech-language pathologists, neurologists, and other clinicians and researchers can lead to different diagnostic labels that may evolve throughout disease progression. These discrepancies and changes can complicate patient understanding, clinical care, and affect eligibility for clinical trials. We highlight the importance of diagnostic specificity, flexibility, and multidisciplinary collaboration in improving patient counseling, prognostication, treatment planning, and clinical research alignment, particularly in the context of the growing number of clinical trials and specialized, multidisciplinary clinics. Conclusions: This viewpoint explores how symptom-based, syndrome-based, and pathology-based diagnoses can work together to support patient care and research. Accurate and specific diagnosis of speech-language disorders has important implications for patient understanding of their condition, identity, prognostic accuracy, access to care, clinical trial enrollment, and treatment planning and efficacy. Misdiagnosis or vague labeling can result in inappropriate interventions, delayed support, and skewed research findings. We argue that multidisciplinary collaboration and flexibility in recognizing evolving symptom-based and syndrome-based labels over the disease course—and as the fields learn more—is essential to empower patients with accurate, actionable information. Embracing both the specificity and the complexity of progressive speech-language disorders can both enhance communication and facilitate research.

Objective: Progressive apraxia of speech (PAOS) is a neurodegenerative syndrome characterized by impaired motor speech planning and programming, whereas behavioral variant frontotemporal dementia (bvFTD) is characterized by deviant behavioral (e.g., personality and social) features. Clinical and anatomic characteristics of bvFTD in the context of PAOS are understudied. Methods: We identified 12 participants with PAOS and features that were consistent with bvFTD at baseline or follow-up. Eleven completed a head MRI scan. We compared clinical features and anatomical patterns of atrophy in these 11 PAOS-bvFTD participants to 11 matched PAOS participants without bvFTD and 22 age- and sex-matched healthy controls. Statistical Parametric Mapping (SPM) was applied to visualize gray matter volume across both groups compared to controls and each other. Medians and 25th and 75th percentiles were assessed in patients and across groups; Fisher’s Exact Test and Mann–Whitney U tests were applied using BlueSky software, version 10.3.1-Pro. Results: As expected, PAOS-bvFTD participants performed worse on the Frontal Behavioral Inventory (median: 33/72 vs. 10/72), 20-item behavioral assessment scale (4.5/20 vs. 1.0/20), and the Neuropsychiatry inventory (4/36 vs. 1.5/36) compared to the PAOS group (p < 0.01 for all), with no differences in other demographic, neurological, or language tests. Seven of the eleven PAOS-bvFTD participants had bvFTD features develop within three years of symptom onset. The PAOS-bvFTD and PAOS groups showed volume loss in frontal lobe regions compared to controls, with PAOS-bvFTD participants having more prefrontal volume loss than PAOS participants. Conclusions: Behavioral features consistent with bvFTD can co-occur in patients with PAOS and are related to greater atrophy of the prefrontal cortex.