Models Of Speech Production Research Articles

Toward Process-Oriented, Dimensional Approaches for Diagnosis and Treatment of Speech Sound Disorders in Children: Position Statement and Future Perspectives.

Children with speech sound disorders (SSD) form a heterogeneous group, with respect to severity, etiology, proximal causes, speech error characteristics, and response to treatment. Infants develop speech and language in interaction with neurological maturation and general perceptual, motoric, and cognitive skills in a social-emotional context. After a brief introduction into psycholinguistic models of speech production and levels of causation, in this review article, we present an in-depth overview of mechanisms and processes, and the dynamics thereof, which are crucial in typical speech development. These basic mechanisms and processes are: (a) neurophysiological motor refinement, that is, the maturational articulatory mechanisms that drive babbling and the more differentiated production of larger speech patterns; (b) sensorimotor integration, which forms the steering function from phonetics to phonology; and (c) motor hierarchy and articulatory phonology describing the gestural organization of syllables, which underlie fluent speech production. These dynamics have consequences for the diagnosis and further analysis of SSD in children. We argue that current diagnostic classification systems do not do justice to the multilevel, multifactorial, and interactive character of the underlying mechanisms and processes. This is illustrated by a recent Dutch study yielding distinct performance profiles among children with SSD, which allows for a dimensional interpretation of underlying processing deficits. Analyses of mainstream treatments with respect to the treatment goals and the speech mechanisms addressed show that treatment programs are quite transparent in their aims and approach and how they contribute to remediating specific deficits or mechanisms. Recent studies into clinical reasoning reveal that the clinical challenge for speech-language pathologists is how to select the most appropriate treatment at the most appropriate time for each individual child with SSD. We argue that a process-oriented approach has merits as compared to categorical diagnostics as a toolbox to aid in the interpretation of the speech profile in terms of underlying deficits and to connect these to a specific intervention approach and treatment target.

Greek-Cypriot learners’ perception and production of L2 English vowels

Acquiring L2 segments is particularly challenging for L2 learners, especially when the L1 and L2 inventories involve different contrasts and acoustic cues. The present research investigated the perception and production of L2 English vowels by adult Greek-Cypriot learners and examined the effects of orthographic cues and consonantal context in their performance. Perceptual performance was assessed through a forced-choice identification task and production performance through a wordlist-reading and an elicitation task, both analyzed acoustically and through intelligibility ratings. The findings showed the influence of the L1 on both the perception and production of L2 segments, supporting the assumptions of current models of speech perception and production. Learners faced challenges in perceiving the members of L2 contrasts and mostly used their L1 articulatory routines in their productions of L2 vowels. Orthographic cues or consonantal context did not significantly affect learners’ productions or overall perception, although strong contextual effects were observed in individual target vowels. This study is the first to provide an in-depth comparison of CYG and English vowels as well as an examination of the acquisition of L2 English vowels by this population, which can guide EFL teachers and course developers in developing appropriate EFL curricula that incorporate pronunciation instruction.

SELF-ASSESSMENT METACOGNITIVE STRATEGIES IN A SPONTANEOUS ESL SPEECH PRODUCTION CONTEXT

Purpose – Self-assessment is regarded as a complex metacognitive process by scholars. Nevertheless, in the context of English as a Second Language (ESL) speaking, self-assessment practices often rely on assessment criteria and teacher commentaries. However, speaking involves spontaneous expression with limited access to external standards. Therefore, this case study aims to explore the metacognitive strategies that proficient ESL students use during selfassessment in spontaneous speech production contexts. Methodology – Three participants, purposefully selected, participated in two spontaneous group discussions recorded on video. Instances of participants’ dysfluency served as prompts in stimulated recall interviews, complemented by video recordings to validate participants’ responses. The thematic analysis of interview data utilised a conceptual framework integrating O’Malley and Chamot’s (1990) metacognitive strategies and Kormos’ (2006) speech production model. Findings – The analysis revealed that participants employed three key metacognitive strategies – organisational planning, selective attention, and self-monitoring to self-assess their spontaneous speech. Feedback sources, such as their proficiency in the second language (L2) and contextual factors, influenced the application and effectiveness of these strategies during self-assessment (SA). Significance – This study offers insights into how proficient ESL students self-assess their spontaneous speech production, leveraging their available resources. In addition, this study identifies speech challenges the participants encountered and how they applied metacognitive strategies to address them.

Phonological neighbourhood effects in Italian speech production: Evidence from healthy and neurologically impaired populations

ABSTRACT In two speech production studies conducted in Italian, we investigated the impact of phonological neighbourhood properties such as the neighbourhood density and the mean frequency of the neighbours on speech processing. Two populations of healthy (Study 1) and neurologically impaired (Study 2) individuals were tested. We employed multi-regression methods to analyse naming latencies in Study 1 and accuracy rates in Study 2 while controlling for various psycholinguistic predictors. In Study 1, pictures with words from high-density neighbourhoods were named faster than those from low-density neighbourhoods. Additionally, words with high-frequency neighbours were named faster in Study 1 and yielded higher accuracy rates in Study 2. The results suggest facilitatory effects of both the phonological neighbourhood density and frequency neighbourhood variables. Furthermore, we observed interactions between these two phonological neighbourhood variables and name agreement and repetition. Specifically, the facilitation effect was more pronounced for pictures with lower name agreement and during the initial presentation of the pictures. These findings are discussed in the context of previous literature and within the framework of interactive models of speech production.

The Impact of Lexical Bundle Length on L2 Oral Proficiency

Lexical bundles (LBs) are crucial in L2 oral proficiency, yet their complexity in terms of length is under-researched. This study therefore examines the relationship between longer and shorter LBs and oral proficiency among 150 L2 learners of varying proficiency levels at a UK university. Through the analysis of oral presentation data (scores ranging from intermediate to advanced) and employing a combined text-internal and text-external approach (two- to five-word bundles), this study advances an innovative text-internal LB refinement procedure, thus isolating the unique contribution of LB length. Robust regression, dominance analysis, and random forest statistical techniques reveal the predictive power of bigram mutual information (MI) and longer three-to-five-word sequences on higher proficiency scores. Our results show that learners using higher MI score bigrams tend to perform better in their presentations, with a strong positive impact on scores (b = 14.38, 95% CI [8.01, 20.76], t = 4.42; dominance weight = 58.63%). Additionally, the use of longer three-to-five-word phrases also contributes to better performance, though to a lesser extent (dominance weight = 18.80%). These findings highlight the pedagogical potential of a nuanced approach to the strategic deployment of LBs, particularly bigram MI, to foster oral proficiency. Suggestions for future LB proficiency research are discussed in relation to L2 speech production models.

Spectrum estimation for voiced speech using average weighted linear prediction

In many different areas, such as neurophysics, geophysics, and communication, it is vital to effectively handle signals when there is unwanted disturbance. Linear prediction (LP) methods provide precise modeling capabilities for various parametric models and are used in a wide range of fields such as predicting future data, compressing speech, analyzing spectra, filling in missing signals, and decreasing noise. LP is especially beneficial in speech processing systems. Speech analysis and synthesis greatly relies on spectral envelopes, which are closely connected to models of speech production and perception. However, the performance of the LP method may decline when limited information is involved in pitch synchronous analysis. In order to overcome this constraint, we developed the implementation of the Average Weighted Linear Prediction (AWLP) technique, which does not require the use of a window. The AWLP utilizes a signal of weighted prediction error acquired from the autocorrelation process of LP. By performing this action, it provides a more precise estimation of the power spectrum and coefficients used for vocal tract parametric models. Importantly, the effectiveness of the AWLP can be seen in both pitch synchronous and asynchronous analyses, which means it is appropriate for situations where segments are chosen randomly. We confirm the effectiveness of the AWLP technique by employing it on both synthetic voice and real speech. Additionally, we assess its effectiveness by comparing it to the direct autocorrelation technique and a modified autocorrelation (MA) approach. To sum up, the AWLP method presents a hopeful solution for reliable spectral analysis that addresses the divide between pitch synchronous and asynchronous situations.

The encoding of speech modes in motor speech disorders: whispered versus normal speech in apraxia of speech and hypokinetic dysarthria

ABSTRACT Speakers with motor speech disorders (MSD) present challenges in speech production, one of them being the difficulty to adapt their speech to different modes. However, it is unclear whether different types of MSD are similarly affected when it comes to adapting their speech to various communication contexts. This study investigates the encoding of speech modes in individuals with AoS following focal brain damage and in individuals with hypokinetic dysarthria (HD) secondary to Parkinson’s disease. Participants with mild-to-moderate MSD and their age-matched controls performed a delayed production task of pseudo-words in two speech modes: normal and whispered speech. While overall accuracy did not differ significantly across speech modes, participants with AoS exhibited longer response latencies for whispered speech, reflecting difficulties in the initiation of utterances requiring an unvoiced production. In contrast, participants with HD showed faster response latencies for whispered speech, indicating that this speech mode is easier to encode/control for this population. Acoustic durations followed these same trends, with participants with AoS showing greater lengthening for whispered speech as compared to controls and to participants with HD, while participants with HD exhibited milder lengthening. Contrary to the predictions of speech production models, suggesting that speech mode changes might be particularly difficult in dysarthria, the present results suggest that speech mode adaptation rather seems particularly costly for participants with AoS.

Investigating the Correlation between English Spoken Fluency and Working Memory Capacity in both Dialogic and Monologic Presentations

Studies on second language (L2) speech fluency recommend studying fluency in a dialogic context. In response to the researchers' calls, this study introduced monologic and dialogic tasks to investigate the various aspects of speed, breakdown, and repair fluency in the oral performancemost L2 fluency studies have looked at oral fluency in a monologic task. Dialogue is the more authentic and natural way of communication, which is apparent in everyday language use. Currently, there is a scarcity of research examining dialogue fluency in non-native bilingual speakers who share the same L2. The existing body of research on language learning and processing has underscored significant connections between individual differences (IDs) in working memory capacity (WMC) and models of L2 speech production in both first language (L1). Nevertheless, it remains unclear whether variations in WMC are linked to dysfluency in L2 monologue and dialogue. Therefore, this study also aimed to fill this gap in the literature by investigating the correlation between utterance fluency in both monologue and dialogue and WMC. A total of 64 undergraduate Saudi students were given various tasks as part of the study. An argumentative task was presented as a monologue focused on a prevalent topic in the participants’ country. In contrast, during a dialogic discussion task, 32 pairs engaged in exchanging opinions on a popular subject in their country. Additionally, participants underwent two challenging working memory (WM) tests: the Operation Span Test and the Backward Digit Span Test. The findings aligned with prior research, indicating that L2 participants demonstrated greater fluency in dialogue compared to monologue, as evident in speed and breakdown measures of utterance fluency. Interestingly, WMC did not emerge as a robust predictor for variations in L2 oral performances between monologue and dialogue.

Simulation of sentence-level speech with an acoustically driven model of speech production

In a model of speech production developed by Story and Bunton [JASA, 146(4), 2522–2528 (2019)] speech segments are encoded by specifying relative acoustic events along a time axis that consist of directional changes of the vocal tract resonance frequencies called resonance deflection patterns (RDPs). These events are transformed via acoustic sensitivity functions, into time-varying modulations of the vocal tract shape. For example, RDPs specifying /b, p/, /d, t/, and /g, k/ would typically be coded as [−1 −1 −1], [−1 1 1], and [−1 1 −1], respectively. The RDP changes, indicate, from left to right, the targeted directional shift of the first, second, and third resonances of the vocal tract. In addition, events that produce nasalization, voiced versus voiceless sounds, and changes in fundamental frequency are also specified the time axis. The purpose of this study was to demonstrate the use of this model to generate word- and sentence-level speech for eventual speech intelligibility studies. The process for simulating several words and sentences will be demonstrated.

Investigation of machine-learning-based stimuli for the remediation of children’s speech errors

Children with speech sound disorders (SSDs) demonstrate difficulty producing phonemes correctly and may exhibit poor speech perception compared to their age-matched peers; however, group differences in speech perception skills remain largely unexplained. Developmental models of speech production posit that children’s ability to discriminate correct and incorrect productions in their own speech may be critical for developing accurate speech production. Historically, when children regularly mispronounce a phoneme, it has been essentially impossible to assess whether they perceive correct versus errors productions in their own speech, thus creating a clinical conundrum. How can we assess a child’s ability to perceive the accuracy of their own phoneme production when they cannot produce a correct production? Recent technological developments allow for acoustic alteration of children’s speech that digitally corrects speech sound errors while preserving natural characteristics of the child’s voice. This machine-learning-based stimuli may then be used as training and feedback tokens for remediation when treating children with speech sound disorders. Such acoustic alteration is possible within an accessible, user-friendly environment that is clinically feasible for speech-language pathologists with little acoustic training. Thus, the purpose of this study is to evaluate the acoustic and perceptual accuracy of acoustically-altered child-speech compared to natural speech tokens.

Beatboxing: Teaching technical acoustical concepts through pop culture

The motivation for this presentation is to demonstrate that vocal percussion, or “beatboxing,” can be a rich method of teaching and demonstrating various introductory acoustical concepts. Beatboxing can be an entertaining way to demonstrate the different acoustical sources of energy—namely burst, noise, and voice—and how they can be combined to create familiar and unfamiliar phonemes, such as the “PF” snare. It also provides an engaging way to introduce the anatomy of the vocal tract as the students can see and hear how vibrating different parts of the mouth and throat (e.g., vocal cords, epiglottis, vestibular folds, lips, etc.) alter the sound. Finally, beatboxing can be used to demonstrate the source-filter model of speech production through altering the sound source in the throat and displaying how to change your mouth shape to produce different formant-altering effects, such as the rapid formant shifting that gives a record scratch its characteristic sound. Beatboxing provides an entertaining and effective way to demonstrate technical concepts to students in an artistic format which captures the attention, especially of younger audiences. [Undergraduate research supported by the College of Physical and Mathematical Sciences, at Brigham Young University.]

Optimization-based modeling of Lombard speech articulation: Supraglottal characteristics.

This paper shows that a highly simplified model of speech production based on the optimization of articulatory effort versus intelligibility can account for some observed articulatory consequences of signal-to-noise ratio. Simulations of static vowels in the presence of various background noise levels show that the model predicts articulatory and acoustic modifications of the type observed in Lombard speech. These features were obtained only when the constraint applied to articulatory effort decreases as the level of background noise increases. These results support the hypothesis that Lombard speech is listener oriented and speakers adapt their articulation in noisy environments.

The relation of velopharyngeal coupling area and vocal tract scaling to identification of stop-nasal cognates.

The purpose of this study was to determine whether the threshold of velopharyngeal (VP) coupling area at which listeners switch from identifying a consonant as a stop to a nasal in North American English was different for speech produced by a model based on an adult male, an adult female, and a 4-year-old child. V1CV2 stimuli were generated with a speech production model that encodes phonetic segments as relative acoustic targets imposed on an underlying vocal tract and laryngeal structure that can be scaled according to sex and age. Each V1CV2 was synthesized with a set of VP coupling functions whose maximum area ranged from 0 to 0.1 cm2. Results showed that scaling the vocal tract and vocal folds had essentially no effect on the VP coupling area at which listener identification shifted from stop to nasal. The range of coupling areas at which the crossover occurred was 0.037-0.049 cm2 for the male model, 0.040-0.055 cm2 for the female model, and 0.039-0.052 cm2 for the 4-year-old child model, and overall mean was 0.044 cm2. Calculations of band limited peak nasalance indicated that 85% peak nasalance during the consonant was well aligned with listener responses.

The impact of response congruence on speech production: An event-related potentials study

A puzzling finding in the speech production literature is the facilitation of categorically related distractors in a superordinate level naming task. The context is in this case response congruent, because application of the task instruction to the context would lead to the correct response. This study investigates the time-course of response congruence effects in speech production using event-related potentials (ERPs). Participants overtly named target words that were overlaid on context pictures with either their superordinate category level name or their associated function, while their response times and ERPs were recorded. Behavioural results replicate the facilitating effect of response congruence. The ERP results showed that the N2 was larger for a response incongruent than congruent context, and this effect correlated with the behavioural pattern of results. This key finding suggests that response incongruence is associated with a conflict-monitoring response which drives the behavioural effect. Further, N400 amplitude was not modulated by response congruence, showing that its effect appears confined to the conceptualisation phase. Finally, P3 modulations mirrored those in RTs, but unlike the N2 effect, they did not correlate with RTs. This suggests that, although the facilitating effect of response congruence is confined to the conceptualisation phase of speech production, response incongruent representations may remain active during later processing stages, or that this late effect of response congruence reflects conflict resolve. Implications for models of speech production are discussed.

Automatic segmentation of vocal tract articulators in real-time magnetic resonance imaging

Background and ObjectivesThe characterization of the vocal tract geometry during speech interests various research topics, including speech production modeling, motor control analysis, and speech therapy design. Real-time MRI is a reliable and non-invasive tool for this purpose. In most cases, it is necessary to know the contours of the individual articulators from the glottis to the lips. Several techniques have been proposed for segmenting vocal tract articulators, but most are limited to specific applications. Moreover, they often do not provide individualized contours for all soft-tissue articulators in a multi-speaker configuration. MethodsA Mask R-CNN network was trained to detect and segment the vocal tract articulator contours in two real-time MRI (RT-MRI) datasets with speech recordings of multiple speakers. Two post-processing algorithms were then proposed to convert the network's outputs into geometrical curves. Nine articulators were considered: the two lips, tongue, soft palate, pharynx, arytenoid cartilage, epiglottis, thyroid cartilage, and vocal folds. A leave-one-out cross-validation protocol was used to evaluate inter-speaker generalization. The evaluation metrics were the point-to-closest-point distance and the Jaccard index (for articulators annotated as closed contours). ResultsThe proposed method accurately segmented the vocal tract articulators, with an average root mean square point-to-closest-point distance of less than 2.2mm for all the articulators in the leave-one-out cross-validation setting. The minimum P2CPRMS was 0.91mm for the upper lip, and the maximum was 2.18mm for the tongue. The Jaccard indices for the thyroid cartilage and vocal folds were 0.60 and 0.61, respectively. Additionally, the method adapted to a new subject with only ten annotated samples. ConclusionsOur research introduced a method for individually segmenting nine non-rigid vocal tract articulators in real-time MRI movies. The software is openly available as an installable package to the speech community. It is designed to develop speech applications and clinical and non-clinical research in fields that require vocal tract geometry, such as speech, singing, and human beatboxing.

How do EFL learners process oral tasks with different complexity: an exploratory study.

The effects of task complexity on learners' performance has been a much-researched issue in SLA field. However, until now many studies fail to provide empirical evidence of the effects of task complexity on learners' processing. To fill the gap, this study examined how task complexity affects L2 learners' cognitive processes with reference to Levelt's speech production model (1989). Ten Chinese EFL learners were asked to complete two narrative tasks with different complexity manipulated by +/- few elements under the same planning conditions. Results revealed that: (1) in the complex task learners showed a slightly lower percentage of cognitive processes at the stage of conceptualization and formulation and a higher percentage linked to monitoring at the stage of comprehension; (2) learners' fluency in oral performance was affected by the cognitive processes at all the stages of oral production. Accuracy seemed to be most enhanced by learners' form monitoring at the comprehension stage. The study sheds light on how learners process the tasks with different complexity when producing the language, and is of implication to task-based language teaching.

Modeling reduced speech using a simulated vocal tract and overlapping acoustic events

Reduced utterances are characterized by the alteration or deletion of segments (i.e., syllables, words) and often occur in spontaneous or casual speech styles. These reductions are common in everyday communication and are generally understood given sufficient context. However, modeling reduced speech is not trivial, as the effects of context, articulatory limitations, idiosyncrasies, and speech rate need to be considered in order to approximate acoustic observations of reduction. We utilize a speech production model which modulates a simulated vocal tract according to acoustic events organized along a time axis. The present study focuses on validating its capacity for modeling reduced speech. Evidence of its effectiveness is provided by way of synthetic example utterances, each accompanied by a description of the steps taken to synthesize the utterance and acoustic comparison to an unreduced equivalent. Word-medial consonant reductions will be studied first before proceeding to massive reductions where entire syllables and words are affected. We iteratively synthesize reduced utterances until they are evaluated through informal listening to be comparable to real acoustic observations. The results of this study are discussed as validation of the idea that acoustic events can be arranged to model reductions.

Semantic processing during continuous speech production: an analysis from eye movements and EEG.

Speech production involves neurological planning and articulatory execution. How speakers prepare for articulation is a significant aspect of speech production research. Previous studies have focused on isolated words or short phrases to explore speech planning mechanisms linked to articulatory behaviors, including investigating the eye-voice span (EVS) during text reading. However, these experimental paradigms lack real-world speech process replication. Additionally, our understanding of the neurological dimension of speech planning remains limited. This study examines speech planning mechanisms during continuous speech production by analyzing behavioral (eye movement and speech) and neurophysiological (EEG) data within a continuous speech production task. The study specifically investigates the influence of semantic consistency on speech planning and the occurrence of "look ahead" behavior. The outcomes reveal the pivotal role of semantic coherence in facilitating fluent speech production. Speakers access lexical representations and phonological information before initiating speech, emphasizing the significance of semantic processing in speech planning. Behaviorally, the EVS decreases progressively during continuous reading of regular sentences, with a slight increase for non-regular sentences. Moreover, eye movement pattern analysis identifies two distinct speech production modes, highlighting the importance of semantic comprehension and prediction in higher-level lexical processing. Neurologically, the dual pathway model of speech production is supported, indicating a dorsal information flow and frontal lobe involvement. The brain network linked to semantic understanding exhibits a negative correlation with semantic coherence, with significant activation during semantic incoherence and suppression in regular sentences. The study's findings enhance comprehension of speech planning mechanisms and offer insights into the role of semantic coherence in continuous speech production. Furthermore, the research methodology establishes a valuable framework for future investigations in this domain.

Cortico-Cerebellar Monitoring of Speech Sequence Production.

In a functional magnetic resonance imaging study, we examined speech error monitoring in a cortico-cerebellar network for two contrasts: (a) correct trials with high versus low articulatory error probability and (b) overtly committed errors versus correct trials. Engagement of the cognitive cerebellar region Crus I in both contrasts suggests that this region is involved in overarching performance monitoring. The activation of cerebellar motor regions (superior medial cerebellum, lobules VI and VIII) indicates the additional presence of a sensorimotor driven implementation of control. The combined pattern of pre-supplementary motor area (active across contrasts) and anterior cingulate cortex (only active in the contrast involving overt errors) activations suggests sensorimotor driven feedback monitoring in the medial frontal cortex, making use of proprioception and auditory feedback through overt errors. Differential temporal and parietal cortex activation across contrasts indicates involvement beyond sensorimotor driven feedback in line with speech production models that link these regions to auditory target processing and internal modeling-like mechanisms. These results highlight the presence of multiple, possibly hierarchically interdependent, mechanisms that support the optimizing of speech production.

The NEF-SPA Approach as a Framework for Developing a Neurobiologically Inspired Spiking Neural Network Model for Speech Production.

The computer-based simulation of the whole processing route for speech production and speech perception in a neurobiologically inspired way remains a challenge. Only a few neural based models of speech production exist, and these models either concentrate on the cognitive-linguistic component or the lower-level sensorimotor component of speech production and speech perception. Moreover, these existing models are second-generation neural network models using rate-based neuron approaches. The aim of this paper is to describe recent work developing a third-generation spiking-neuron neural network capable of modeling the whole process of speech production, including cognitive and sensorimotor components. Our neural model of speech production was developed within the Neural Engineering Framework (NEF), incorporating the concept of Semantic Pointer Architecture (SPA), which allows the construction of large-scale neural models of the functioning brain based on only a few essential and neurobiologically well-grounded modeling or construction elements (i.e., single spiking neuron elements, neural connections, neuron ensembles, state buffers, associative memories, modules for binding and unbinding of states, modules for time scale generation (oscillators) and ramp signal generation (integrators), modules for input signal processing, modules for action selection, etc.). We demonstrated that this modeling approach is capable of constructing a fully functional model of speech production based on these modeling elements (i.e., biologically motivated spiking neuron micro-circuits or micro-networks). The model is capable of (i) modeling the whole processing chain of speech production and, in part, for speech perception based on leaky-integrate-and-fire spiking neurons and (ii) simulating (macroscopic) speaking behavior in a realistic way, by using neurobiologically plausible (microscopic) neural construction elements. The model presented here is a promising approach for describing speech processing in a bottom-up manner based on a set of micro-circuit neural network elements for generating a large-scale neural network. In addition, the model conforms to a top-down design, as it is available in a condensed form in box-and-arrow models based on functional imaging and electrophysiological data recruited from speech processing tasks.