Are there prototypical associations between time frames and aspectual values? Evidence from Greek aphasia and healthy ageing.

Time reference, which has been found to be selectively impaired in agrammatic aphasia, is often interwoven with grammatical aspect. A recent study on Russian aphasia found that time reference and aspect interact: Past reference was less impaired when tested within a perfective aspect context (compared to when tested within an imperfective aspect context), and reference to the non-past was less impaired when tested within an imperfective aspect context (compared to when tested within a perfective aspect context). To explain this pattern, the authors argued that there are prototypical associations between time frames and aspectual values. The present study explores the relationship between time reference and aspect focusing on Greek aphasia and healthy ageing and using a sentence completion task that crosses time reference and aspect. The findings do not support prototypical matches between different time frames and aspectual values. Building on relevant studies, we propose that patterns of performance of healthy or language-impaired speakers on constrained tasks tapping different combinations of time frames with aspectual values should reflect the relative frequency of these combinations in a given language. The analysis of the results at the individual level revealed a double dissociation, which indicates that a given time frame-aspectual value combination may be relatively easy to process for some persons with aphasia but demanding for some others.

The Nature of Phonetic Gradience across a Dialect Continuum: Evidence from Modern Greek Vowels.

This study investigates the acoustic properties of vowels in 2 Modern Greek varieties: Standard Modern Greek (SMG) and Cypriot Greek (CG). Both varieties contain in their phonetic inventories the same 5 vowels. Forty-five female speakers between 19 and 29 years old participated in this study: 20 SMG speakers and 25 CG speakers, born and raised in Athens and Nicosia, respectively. Stimuli consisted of a set of nonsense CVCV and VCV words, each containing 1 of the 5 Greek vowels in stressed and unstressed position. Gaining insights from the controlled experimental design, the study sheds light on the gradient effects of vowel variation in Modern Greek. It shows that (1) stressed vowels are more peripheral than unstressed vowels, (2) SMG unstressed /i a u/ vowels are more raised than the corresponding CG vowels, (3) SMG unstressed vowels are shorter than CG unstressed vowels, and (4) SMG /i·u/ are more rounded than the corresponding CG vowels. Moreover, it shows that variation applies to specific subsystems, as it is the unstressed vowels that vary cross-varietally whereas the stressed vowels display only minor differences. The implications of these findings with respect to vowel raising and vowel reduction are discussed.

The bursts of stops can convey dialectal information.

This study investigates the effects of the dialect of the speaker on the spectral properties of stop bursts. Forty-five female speakers-20 Standard Modern Greek and 25 Cypriot Greek speakers-participated in this study. The spectral properties of stop bursts were calculated from the burst spectra and analyzed using spectral moments. The findings show that besides linguistic information, i.e., the place of articulation and the stress, the speech signals of bursts can encode social information, i.e., the dialects. A classification model using decision trees showed that skewness and standard deviation have a major contribution for the classification of bursts across dialects.

Seeking an Anchorage. Stability and Variability in Tonal Alignment of Rising Prenuclear Pitch Accents in Cypriot Greek.

Although tonal alignment constitutes a quintessential property of pitch accents, its exact characteristics remain unclear. This study, by exploring the timing of the Cypriot Greek L*+H prenuclear pitch accent, examines the predictions of three hypotheses about tonal alignment: the invariance hypothesis, the segmental anchoring hypothesis, and the segmental anchorage hypothesis. The study reports on two experiments: the first of which manipulates the syllable patterns of the stressed syllable, and the second of which modifies the distance of the L*+H from the following pitch accent. The findings on the alignment of the low tone (L) are illustrative of the segmental anchoring hypothesis predictions: the L persistently aligns inside the onset consonant, a few milliseconds before the stressed vowel. However, the findings on the alignment of the high tone (H) are both intriguing and unexpected: the alignment of the H depends on the number of unstressed syllables that follow the prenuclear pitch accent. The 'wandering' of the H over multiple syllables is extremely rare among languages, and casts doubt on the invariance hypothesis and the segmental anchoring hypothesis, as well as indicating the need for a modified version of the segmental anchorage hypothesis. To address the alignment of the H, we suggest that it aligns within a segmental anchorage-the area that follows the prenuclear pitch accent-in such a way as to protect the paradigmatic contrast between the L*+H prenuclear pitch accent and the L+H* nuclear pitch accent.

Machine Learning Classification of Patients with Amnestic Mild Cognitive Impairment and Non-Amnestic Mild Cognitive Impairment from Written Picture Description Tasks.

Individuals with Mild Cognitive Impairment (MCI), a transitional stage between cognitively healthy aging and dementia, are characterized by subtle neurocognitive changes. Clinically, they can be grouped into two main variants, namely patients with amnestic MCI (aMCI) and non-amnestic MCI (naMCI). The distinction of the two variants is known to be clinically significant as they exhibit different progression rates to dementia. However, it has been particularly challenging to classify the two variants robustly. Recent research indicates that linguistic changes may manifest as one of the early indicators of pathology. Therefore, we focused on MCI's discourse-level writing samples in this study. We hypothesized that a written picture description task can provide information that can be used as an ecological, cost-effective classification system between the two variants. We included one hundred sixty-nine individuals diagnosed with either aMCI or naMCI who received neurophysiological evaluations in addition to a short, written picture description task. Natural Language Processing (NLP) and a BERT pre-trained language model were utilized to analyze the writing samples. We showed that the written picture description task provided 90% overall classification accuracy for the best classification models, which performed better than cognitive measures. Written discourses analyzed by AI models can automatically assess individuals with aMCI and naMCI and facilitate diagnosis, prognosis, therapy planning, and evaluation.

Autism Detection in Children: Integrating Machine Learning and Natural Language Processing in Narrative Analysis.

Despite the consensus that early identification leads to better outcomes for individuals with autism spectrum disorder (ASD), recent research reveals that the average age of diagnosis in the Greek population is approximately six years. However, this age of diagnosis is delayed by an additional two years for families from lower-income or minority backgrounds. These disparities result in adverse impacts on intervention outcomes, which are further burdened by the often time-consuming and labor-intensive language assessments for children with ASD. There is a crucial need for tools that increase access to early assessment and diagnosis that will be rigorous and objective. The current study leverages the capabilities of artificial intelligence to develop a reliable and practical model for distinguishing children with ASD from typically-developing peers based on their narrative and vocabulary skills. We applied natural language processing-based extraction techniques to automatically acquire language features (narrative and vocabulary skills) from storytelling in 68 children with ASD and 52 typically-developing children, and then trained machine learning models on the children's combined narrative and expressive vocabulary data to generate behavioral targets that effectively differentiate ASD from typically-developing children. According to the findings, the model could distinguish ASD from typically-developing children, achieving an accuracy of 96%. Specifically, out of the models used, hist gradient boosting and XGBoost showed slightly superior performance compared to the decision trees and gradient boosting models, particularly regarding accuracy and F1 score. These results bode well for the deployment of machine learning technology for children with ASD, especially those with limited access to early identification services.

The Contribution of Working Memory Areas to Verbal Learning and Recall in Primary Progressive Aphasia.

Recent evidence of domain-specific working memory (WM) systems has identified the areas and networks which are involved in phonological, orthographic, and semantic WM, as well as in higher level domain-general WM functions. The contribution of these areas throughout the process of verbal learning and recall is still unclear. In the present study, we asked, what is the contribution of domain-specific specialized WM systems in the course of verbal learning and recall? To answer this question, we regressed the perfusion data from pseudo-continuous arterial spin labeling (pCASL) MRI with all the immediate, consecutive, and delayed recall stages of the Rey Auditory Verbal Learning Test (RAVLT) from a group of patients with Primary Progressive Aphasia (PPA), a neurodegenerative syndrome in which language is the primary deficit. We found that the early stages of verbal learning involve the areas with subserving phonological processing (left superior temporal gyrus), as well as semantic WM memory (left angular gyrus, AG_L). As learning unfolds, areas with subserving semantic WM (AG_L), as well as lexical/semantic (inferior temporal and fusiform gyri, temporal pole), and episodic memory (hippocampal complex) become more involved. Finally, a delayed recall depends entirely on semantic and episodic memory areas (hippocampal complex, temporal pole, and gyri). Our results suggest that AG_L subserving domain-specific (semantic) WM is involved only during verbal learning, but a delayed recall depends only on medial and cortical temporal areas.

Effects of tDCS on Sound Duration in Patients with Apraxia of Speech in Primary Progressive Aphasia.

Transcranial direct current stimulation (tDCS) over the left inferior frontal gyrus (IFG) was found to improve oral and written naming in post-stroke and primary progressive aphasia (PPA), speech fluency in stuttering, a developmental speech-motor disorder, and apraxia of speech (AOS) symptoms in post-stroke aphasia. This paper addressed the question of whether tDCS over the left IFG coupled with speech therapy may improve sound duration in patients with apraxia of speech (AOS) symptoms in non-fluent PPA (nfvPPA/AOS) more than sham. Eight patients with non-fluent PPA/AOS received either active or sham tDCS, along with speech therapy for 15 sessions. Speech therapy involved repeating words of increasing syllable-length. Evaluations took place before, immediately after, and two months post-intervention. Words were segmented into vowels and consonants and the duration of each vowel and consonant was measured. Segmental duration was significantly shorter after tDCS compared to sham and tDCS gains generalized to untrained words. The effects of tDCS sustained over two months post-treatment in trained and untrained sounds. Taken together, these results demonstrate that tDCS over the left IFG may facilitate speech production by reducing segmental duration. The results provide preliminary evidence that tDCS may maximize efficacy of speech therapy in patients with nfvPPA/AOS.

Part of Speech Production in Patients With Primary Progressive Aphasia: An Analysis Based on Natural Language Processing.

Background Primary progressive aphasia (PPA) is a neurodegenerative disorder characterized by a progressive decline of language functions. Its symptoms are grouped into three PPA variants: nonfluent PPA, logopenic PPA, and semantic PPA. Grammatical deficiencies differ depending on the PPA variant. Aims This study aims to determine the differences between PPA variants with respect to part of speech (POS) production and to identify morphological markers that classify PPA variants using machine learning. By fulfilling these aims, the overarching goal is to provide objective measures that can facilitate clinical diagnosis, evaluation, and prognosis. Method and Procedure Connected speech productions from PPA patients produced in a picture description task were transcribed, and the POS class of each word was estimated using natural language processing, namely, POS tagging. We then implemented a twofold analysis: (a) linear regression to determine how patients with nonfluent PPA, semantic PPA, and logopenic PPA variants differ in their POS productions and (b) a supervised classification analysis based on POS using machine learning models (i.e., random forests, decision trees, and support vector machines) to subtype PPA variants and generate feature importance (FI). Outcome and Results Using an automated analysis of a short picture description task, this study showed that content versus function words can distinguish patients with nonfluent PPA, semantic PPA, and logopenic PPA variants. Verbs were less important as distinguishing features of patients with different PPA variants than earlier thought. Finally, the study showed that among the most important distinguishing features of PPA variants were elaborative speech elements, such as adjectives and adverbs.

Automatic Subtyping of Individuals with Primary Progressive Aphasia.

BACKGROUND: The classification of patients with primary progressive aphasia (PPA) into variants is time-consuming, costly, and requires combined expertise by clinical neurologists, neuropsychologists, speech pathologists, and radiologists. OBJECTIVE: The aim of the present study is to determine whether acoustic and linguistic variables provide accurate classification of PPA patients into one of three variants: nonfluent PPA, semantic PPA, and logopenic PPA. METHODS: In this paper, we present a machine learning model based on deep neural networks (DNN) for the subtyping of patients with PPA into three main variants, using combined acoustic and linguistic information elicited automatically via acoustic and linguistic analysis. The performance of the DNN was compared to the classification accuracy of Random Forests, Support Vector Machines, and Decision Trees, as well as to expert clinicians' classifications. RESULTS: The DNN model outperformed the other machine learning models as well as expert clinicians' classifications with 80% classification accuracy. Importantly, 90% of patients with nfvPPA and 95% of patients with lvPPA was identified correctly, providing reliable subtyping of these patients into their corresponding PPA variants. CONCLUSION: We show that the combined speech and language markers from connected speech productions can inform variant subtyping in patients with PPA. The end-to-end automated machine learning approach we present can enable clinicians and researchers to provide an easy, quick, and inexpensive classification of patients with PPA.

Standardizing Assessment of Spoken Discourse in Aphasia: A Working Group With Deliverables.

Purpose The heterogeneous nature of measures, methods, and analyses reported in the aphasia spoken discourse literature precludes comparison of outcomes across studies (e.g., meta-analyses) and inhibits replication. Furthermore, funding and time constraints significantly hinder collecting test-retest data on spoken discourse outcomes. This research note describes the development and structure of a working group, designed to address major gaps in the spoken discourse aphasia literature, including a lack of standardization in methodology, analysis, and reporting, as well as nominal data regarding the psychometric properties of spoken discourse outcomes. Method The initial initiatives for this working group are to (a) propose recommendations regarding standardization of spoken discourse collection, analysis, and reporting in aphasia, based on the results of an international survey and a systematic literature review and (b) create a database of test-retest spoken discourse data from individuals with and without aphasia. The survey of spoken discourse collection, analysis, and interpretation procedures was distributed to clinicians and researchers involved in aphasia assessment and rehabilitation from September to November 2019. We will publish survey results and recommend standards for collecting, analyzing, and reporting spoken discourse in aphasia. A multisite endeavor to collect test-retest spoken discourse data from individuals with and without aphasia will be initiated. This test-retest information will be contributed to a central site for transcription and analysis, and data will be subsequently openly curated. Conclusion The goal of the working group is to create recommendations for field-wide standards in methods, analysis, and reporting of spoken discourse outcomes, as has been done across other related disciplines (e.g., Consolidated Standards of Reporting Trials, Enhancing the Quality and Transparency of Health Research, Committee on Best Practice in Data Analysis and Sharing). Additionally, the creation of a database through our multisite collaboration will allow the identification of psychometrically sound outcome measures and norms that can be used by clinicians and researchers to assess spoken discourse abilities in aphasia.

Voice quality and speech fluency distinguish individuals with Mild Cognitive Impairment from Healthy Controls.

Mild Cognitive Impairment (MCI) is a syndrome characterized by cognitive decline greater than expected for an individual's age and education level. This study aims to determine whether voice quality and speech fluency distinguish patients with MCI from healthy individuals to improve diagnosis of patients with MCI. We analyzed recordings of the Cookie Theft picture description task produced by 26 patients with MCI and 29 healthy controls from Sweden and calculated measures of voice quality and speech fluency. The results show that patients with MCI differ significantly from HC with respect to acoustic aspects of voice quality, namely H1-A3, cepstral peak prominence, center of gravity, and shimmer; and speech fluency, namely articulation rate and averaged speaking time. The method proposed along with the obtainability of connected speech productions can enable quick and easy analysis of speech fluency and voice quality, providing accessible and objective diagnostic markers of patients with MCI.

A Tool for Automatic Scoring of Spelling Performance.

Purpose The evaluation of spelling performance in aphasia reveals deficits in written language and can facilitate the design of targeted writing treatments. Nevertheless, manual scoring of spelling performance is time-consuming, laborious, and error prone. We propose a novel method based on the use of distance metrics to automatically score spelling. This study compares six automatic distance metrics to identify the metric that best corresponds to the gold standard-manual scoring-using data from manually obtained spelling scores from individuals with primary progressive aphasia. Method Three thousand five hundred forty word and nonword spelling productions from 42 individuals with primary progressive aphasia were scored manually. The gold standard-the manual scores-were compared to scores from six automated distance metrics: sequence matcher ratio, Damerau-Levenshtein distance, normalized Damerau-Levenshtein distance, Jaccard distance, Masi distance, and Jaro-Winkler similarity distance. We evaluated each distance metric based on its correlation with the manual spelling score. Results All automatic distance scores had high correlation with the manual method for both words and nonwords. The normalized Damerau-Levenshtein distance provided the highest correlation with the manual scoring for both words (rs = .99) and nonwords (rs = .95). Conclusions The high correlation between the automated and manual methods suggests that automatic spelling scoring constitutes a quick and objective approach that can reliably substitute the existing manual and time-consuming spelling scoring process, an important asset for both researchers and clinicians.

Identification of Mild Cognitive Impairment From Speech in Swedish Using Deep Sequential Neural Networks.

While people with mild cognitive impairment (MCI) portray noticeably incipient memory difficulty in remembering events and situations along with problems in decision making, planning, and finding their way in familiar environments, detailed neuropsychological assessments also indicate deficits in language performance. To this day, there is no cure for dementia but early-stage treatment can delay the progression of MCI; thus, the development of valid tools for identifying early cognitive changes is of great importance. In this study, we provide an automated machine learning method, using Deep Neural Network Architectures, that aims to identify MCI. Speech materials were obtained using a reading task during evaluation sessions, as part of the Gothenburg MCI research study. Measures of vowel duration, vowel formants (F1 to F5), and fundamental frequency were calculated from speech signals. To learn the acoustic characteristics associated with MCI vs. healthy controls, we have trained and evaluated ten Deep Neural Network Architectures and measured how accurately they can diagnose participants that are unknown to the model. We evaluated the models using two evaluation tasks: a 5-fold crossvalidation and by splitting the data into 90% training and 10% evaluation set. The findings suggest first, that the acoustic features provide significant information for the identification of MCI; second, the best Deep Neural Network Architectures can classify MCI and healthy controls with high classification accuracy (M = 83%); and third, the model has the potential to offer higher accuracy than 84% if trained with more data (cf., SD≈15%). The Deep Neural Network Architecture proposed here constitutes a method that contributes to the early diagnosis of cognitive decline, quantify the progression of the condition, and enable suitable therapeutics.

Effects of Two Linguistically Proximal Varieties on the Spectral and Coarticulatory Properties of Fricatives: Evidence from Athenian Greek and Cypriot Greek.

Several studies have explored the acoustic structure of fricatives, yet there has been very little acoustic research on the effects of dialects on the production of fricatives. This article investigates the effects of two linguistically proximal Modern Greek dialects, Athenian Greek and Cypriot Greek on the temporal, spectral, and coarticulatory properties of fricatives and aims to determine the acoustic properties that convey information about these two dialects. Productions of voiced and voiceless labiodental, dental, alveolar, palatal, and velar fricatives were extracted from a speaking task from typically speaking female adult speakers (25 Cypriot Greek and 20 Athenian Greek speakers). Measures were made of spectral properties, using a spectral moments analysis. The formants of the following vowel were measured and second degree polynomials of the formant contours were calculated. The findings showed that Athenian Greek and Cypriot Greek fricatives differ in all spectral properties across all places of articulation. Also, the co-articulatory effects of fricatives on following vowel were different depending on the dialect. Duration, spectral moments, and the starting frequencies of F1, F2, F3, and F4 contributed the most to the classification of dialect. These findings provide a solid evidence base for the manifestation of dialectal information in the acoustic structure of fricatives.