AbstractParkinson’s disease (PD), a multifaceted neurodegenerative disorder, can manifest as an array of motor and non-motor symptoms. Among these, speech and language impairments are particularly prevalent, often preceding motor dysfunctions. Emerging research indicates that these impairments may serve as early disease indicators. In this narrative review, we synthesised current findings on the potential of speech and language symptoms in PD identification and progression monitoring. Our review highlights convergent, albeit preliminary, lines of evidence supporting the value of speech-related features in detecting early or prodromal PD, even across language groups, especially with sophisticated analytical techniques. Distinct speech patterns in PD subtypes and other neurological disorders may assist in differential diagnosis and inform targeted management efforts. These features also evolve over the disease course and could effectively be utilised for disease tracking and guide management plan modifications. Advances in digital voice processing allow cost-effective, remote and scalable monitoring for larger populations.
IntroductionParkinson’s disease (PD) is a complex neurodegenerative disorder characterised by intracellular deposits of α-synuclein and progressive loss of dopaminergic neurons in the substantia nigra pars compacta1. It is the fastest-growing and second-most common neurological disease currently affecting approximately 10 million individuals worldwide with the figure expected to double by 20502. While the principal manifestations include impaired motor functions (e.g., resting tremor, rigidity and bradykinesia), a multitude of non-motor symptoms occur concomitantly throughout the disease course1,3,4, further contributing to the overall disease burden and poorer quality of life for people with PD (PwPD) and their caregivers and loved ones.There is a recent shift to a biological definition of PD integrating pathogenesis, genetics, biomarkers and clinical components of both motor and non-motor features5,6; however, these newly proposed criteria are intended exclusively for research purposes. Clinically, a diagnosis still relies primarily on motor dysfunctions, specifically the presence of bradykinesia combined with rigidity and/or resting tremor1. By this point, ~50–60% of substantial nigra dopaminergic neuron loss and up to 70% of striatal dopamine depletion have likely occurred7,8, which significantly limits the effectiveness of any neuroprotective or disease-modifying therapies aiming to slow or halt neurodegeneration9,10,11. Sensitive and reliable biomarkers of PD at or before symptom onset are therefore urgently needed for such interventions to be introduced early to achieve the optimal treatment outcome.Rapid eye movement (REM) sleep behaviour disorder (RBD), a parasomnia characterised by dream-enactment behaviours during loss of REM sleep muscle atonia, is considered a key prodromal stage of \(\alpha\)-synucleinopathies12. However, RBD is not a specific or sensitive marker for PD as it also occurs in other neurodegenerative disorders such as dementia with Lewy bodies (DLB) and multiple system atrophy (MSA)12,13, and its prevalence in PD is relatively low ( ~ 24–42%)14,15. Moreover, the need for a polysomnographic confirmation12,16 further restricts its use at the populational level.Speech impairments, on the other hand, are particularly prevalent, affecting up to 90% of PwPD17,18. Growing evidence indicates that speech and language alterations often precede the defining motor signs and PD diagnosis by as much as a decade4,19,20,21,22,23; consequently, several guidelines recommend incorporating speech and language pathology/therapy (SLP) as a crucial component of PD care and rehabilitation from diagnosis24. Recent research highlights the value of objective acoustic speech markers in detecting PD in the initial or even prodromal (e.g., RBD) stages23,25,26. This implicates a potential window for timely interventions, such as monoamine oxidase B inhibitors, anticholinergics and future novel therapies that may confer improved efficacy in early PD stages9,11. Speech and language features might also be used as a proxy measure for tracking disease progression and status as they evolve over the course of the disease25,27,28,29,30,31. Moreover, specific patterns of speech impairment have been linked to the presence of varying disease subtypes characterised by distinct underlying pathologies23,29 and have been shown to differentiate PD from other basal ganglia disorders32 and synucleinopathies23,33,34,35 as well as between PD subtypes36,37.To reflect the current state of knowledge and inform future investigations, we conducted a narrative review synthesising current evidence pertinent to the value of speech and language biomarkers in PD prediction, early detection and progression. Common speech and language difficulties in PD and recommended speech tasks and measurement tools were only briefly described as multiple reviews have addressed these topics3,4,38,39,40,41.MethodsWe searched PubMed, Google Scholar and Web of Science databases using the search terms Parkinson’s disease” OR “Parkinson disease” OR “PD” AND “language” OR “linguistic” OR “speech” OR “acoustics” OR “voice” OR “fluency” OR “intelligibility” OR “communication.” Reference lists from identified publications were also searched. Papers were reviewed and selected based on their relevance to the focus of the review.Speech and language assessmentThe evaluation of speech and linguistic functions encompasses a set of standardised tests that are applicable across various disciplines, including conditions like PD and other movement disorders38,39,40,41. The recommended and widely employed tests are 1) sustained phonation of the vowel /a/ in a single breath, which helps assess breath control, vocal fold function and vocal quality; 2) rapid repetition of syllables such as /pa/, /ta/ and /ka/ to evaluate consonant and vowel articulation, articulation rate and regularity, coordination and speech timing; 3) reading a passage that assesses articulation, prosody and resonance and 4) verbal fluency test (phonemic and semantic naming) to assess lexicon access, verbal functioning and executive control42,43 5) picture description and/or spontaneous monologue, which provides insight into fluency, language formulation at the lexical-semantic, morpho-syntactic and discourse-pragmatic levels and overall speech and language abilities38,39,41.Following the administration of these tests, speech recordings are analysed either perceptually by a qualified speech and language pathologist or objectively through acoustic and linguistic analyses. Perceptual analysis involves the use of standardised protocols to systematically describe speech characteristics (e.g., voice quality, articulation, rate and rhythm), the severity of dysarthria, overall intelligibility and naturalness of speech; these can help gauge the functional aspects of communication deficits17,38,44. Objective analyses, on the other hand, utilise specialised software or digital interfaces to quantify various acoustic parameters related to spectral characteristics of voice signals and linguistic features from speech transcripts, providing a more in-depth and sensitive approach to assessing an individual’s speech and language abilities4,38,41. With automated signal processing, these objective tools are also more accessible and scalable, allowing for remote monitoring.Consideration of self-perception of voice, speech and language impairment is advocated as can provides important insights into the impact of PD on functional communication in real-world contexts45,46,47,48,49. Example tools used in this context include the Voice Handicap Index45, the Dysarthria Impact Scale50 and the Dysarthria Impact Profile48. They provide valuable information on the subjective perception of impact; however, they are less sensitive to minor fluctuations in function and may be affected by flooring and ceiling effects inherent to the evaluation scales employed.Voice, speech and language alterations in PDImpaired speech in PD can result from disruption to some or all subsystems involved in speech production, including phonation, articulation, prosody, respiration and resonance. These impairments are typically characterised by diminished loudness reflected by a soft or asthenic voice51, reduced variability in loudness and pitch resulting in a monopitch and monoloudness25,26,30,51, a harsh and/or breathy voice quality17,32 and imprecise articulation25,52, all contributing to decreased intelligibility17,53. Inappropriate pauses and inconsistent speech rates such as abrupt deceleration or acceleration of speech have also been reported4,26,54,55, which collectively can be classified as “hypokinetic dysarthria”53. These perceptual changes are reflected in objective acoustic features including reduced range and variation in voice intensity and fundamental frequency (f0), shorter maximum phonation time, lower harmonics-to-noise ratio (HNR) and cepstral peak prominence (CPP), restricted vowel space area (VSA) and lower vowel articulation index (VAI), longer voice onset time (VOT), increased perturbations in the amplitude and frequency of sound waves described as jitter and shimmer, respectively, slower and irregular diadochokinetic rate and prolonged pause intervals3,19,30,37,56.While most communication research in PD has focused on motor speech functions, impaired language is associated with PD. The verbal output of PwPD relative to individuals without PD is often less informative23,31,57, concise58, grammatically intact31 and complex58,59, relying more on content words rather than functional words (e.g., prepositions and conjunctions)23 during narrative or other structured tasks. At the conversational level, word-finding difficulties or “tip-of-the-tongue” phenomenon and challenges in initiating and maintaining a topic have been observed4,60. PwPD can also have difficulties recognising and interpreting non-verbal cues such as emotion expressed through facial expression or tones61,62,63, in addition to reduced body gestures and flattened affect3,60,63, making it harder for them to engage in conversations.A short glossary of speech and language terminologies used in this study is outlined in Table 1. A schematic illustrating the strategic integration of speech and language markers across the PD continuum to facilitate clinical decision-making and better management of the disease is depicted in Fig. 1.Table 1 Glossary of speech and language terminologies in this reviewFull size tableFig. 1: Integrating speech and language markers into clinical practice for Parkinson’s disease screening, diagnosis, differential diagnosis, subtyping and monitoring.This schematic demonstrates how integrating speech and language markers into clinical practice across the Parkinson’s disease (PD) continuum can enhance early detection, inform clinical decision-making and optimize disease management throughout its progression.Full size imagePreclinical or prodromal speech and language markers for early PD detection and predictionIn the PD progression trajectory proposed by Braak et al.64, Lewy body pathology in Stages 1–2 is confined to the brainstem, affecting the glossopharyngeal and vagal nerves, which innervate the laryngeal and pharyngeal musculature and coordinate articulatory and respiratory activities19,21,65, and olfactory regions before reaching the substantia nigra. In accordance with Braak’s staging, Selby66 also hypothesised a caudorostral symptom progression in PD, starting with failure in speech-related respiratory control and then articulation involving the larynx, pharynx, tongue and lips. Converging evidence now suggests that voice and speech alterations, along with abnormalities in swallowing and smell, are among the earliest signs of PD and may serve as early markers for diagnosis and prediction.Dysprosody was consistently observed in the early and even prodromal stages of the disease across different languages. In a multicentre cohort study administered in Czech, English, German, French and Italian languages, monopitch differentiated groups, even in the prodromal phase of PD25. In this study, early speech markers were explored in 150 participants with idiopathic RBD, 149 with early-stage PD (mean disease duration 1.7–2.5 years) and 149 controls. Four speech tasks including sustained phonation, fast syllable repetition (/pa//ta//ka/), text reading and narration were administered at baseline and 12-month follow-up. A significant difference was noted in monopitch, prolonged pauses and imprecise consonants between PD and control participants, resulting in an area under the curve (AUC) of 0.80 for group classification. Monopitch and prolonged pauses were also different between RBD and PD, with an AUC of 0.72 when combined with extra features including harsh voice and articulation rate. Similarly, monopitch separated RBD subjects from controls, though with a smaller AUC of 0.65, by incorporating the same additional features.The relationship between prominent speech features like prosody in prodromal PD and other markers of the disease such as altered olfactory functions and nigrostriatal system have also been investigated in Czech67. Similar to earlier work, data suggest that prosodic impairment can present prior to the appearance of nigro-putaminal dopaminergic deficits, with dysprosody and olfactory dysfunctions co-occurring. Reduced pitch variability was detected in people with RBD and hyposmia but not those with normal olfaction compared to healthy individuals67. Additionally, while monopitch at baseline was independent of putamen binding ratio captured by dopamine transporter single-photon emission computed tomography (DAT-SPECT), deterioration in impaired pitch variability was only seen in the RBD subgroup with hyposmia and abnormal DAT-SPECT at the 2-year follow-up.Again, using RBD as a model for prodromal PD, a French-speaking cohort yielded similar outcomes in exploring voice characteristics from prodromal to early PD26. Both RBD and PD groups produced more monotonous speech alongside longer pause durations and an unsteady rhythm, compared to the control group26. Leveraging voice features extracted from automated acoustic analysis and supervised learning classifications, early PD was classified with 89% accuracy for males and 70% for females, while RBD detection achieved an accuracy of 63–70%. It is noteworthy that more pronounced prosodic deficits were observed during emotional encounters during task performance, implying that tasks with higher emotional demands could offer greater sensitivity in detecting early, nuanced speech alterations.Vowel distortion is a key component of dysarthric speech. It is typically measured by exploring the distribution of the first and second formants68. There is some preliminary evidence to suggest composite measures of vowel production differ in PwPD prior to the commencement of therapy and matched controls52. These parameters include the VSA and VAI derived from connected speech tasks. It is also noted that spontaneous speech tasks (e.g., monologue), which are more complex with higher articulatory demands, better separated groups compared to non-spontaneous tasks (e.g., sentence repetition and passage reading)52.Vowel articulation deficits may also be present in the prodromal stage69. In a study exploring whether altered vowel articulation was a prodromal symptom of synucleinopathy in a larger male sample (60 RBD, 60 de-novo (newly diagnosed), untreated PD and 60 age-matched controls), both the RBD and PD groups exhibited significantly smaller VSA during passage reading than controls and lower VAI was found in PD compared to controls69. The results also revealed that imprecise vowels evaluated by both acoustic indices correlated with the Unified Parkinson’s Disease Rating Scale (UPDRS) motor scores and the severity of bradykinesia and rigidity but not axial gait symptoms or tremor. Furthermore, a positive correlation was observed between the putamen binding ratio detected from DAT-SPECT and VSA in RBD, suggesting that nigrostriatal neurodegeneration might play a role in articulatory impairment.Advances in digital voice processing and machine learning utilising speech signals have propelled research forward in the detection and prediction of PD. Sophisticated feature selection and classification models have been proposed to be able to effectively distinguish PwPD from healthy individuals70,71,72,73, regardless of the presence of overt symptoms74, and even people in prodromal PD (e.g., RBD)75 with motor symptoms from those without73, based on sets of simple speech tasks. More recent studies have also demonstrated the scalability and accessibility of these detection tools to a wider population with their capability of operating effectively on mobile devices70,75,76,77. Sustained vowel task, among other tests, was widely adopted65,70,71,72,74. Data suggest that ~99% discrimination accuracy can be achieved65, using even just a handful of dysphonic measures by support vector machines72. However, overfitting is likely due to the model complexity and limited training samples, without validation on novel datasets. When the same technique was applied to a larger cohort (1078 PwPD and 5453 controls), the accuracy dropped to < 70%78, though the reduced performance may also be linked to the poorer sound quality of phone calls78 compared to controlled and standardised speech sampling72.While studies have underscored the value of speech features as early PD markers, especially with advanced analytics, these approaches come with their own set of advantages and challenges. Machine learning algorithms can uncover complex patterns and subtle voice and speech changes in early-stage PD that conventional models might miss, leading to more accurate and nuanced predictions. However, these advanced models often lack interpretability due to their “black box” nature, making it challenging to understand the decision-making processes that are crucial for clinical validation. Additionally, the transferability of these advanced models to clinical settings requires extensive validation to ensure they work across diverse populations and clinical settings.Compared to motor speech features, fewer research efforts have been directed to early PD language and conversational limitations, and that on preclinical or prodromal language alterations remain scarce. However, there is evidence indicating that language deficits are detectable in prodromal PD and tasks with higher cognitive demands can distinguish RBD and PD with mild cognitive impairment (MCI) from controls as well as RBD and PD with intact cognition with an AUC of up to 0.8279. In this study, prodromal linguistic markers were investigated in 40 RBD cases and 40 de-novo PD cases without MCI, 14 RBD cases and 15 de-novo PD cases with MCI and 30 controls79. It is observed that both RBD groups had lower content density than controls during spontaneous discourse, whereas RBD and PD subgroups with MCI exhibited poorer lexical diversity and more frequent phrase repetitions than both subgroups without MCI and controls in a story narration task. The results also showed that only 7.5% of RBD subjects without MCI had an abnormal DAT-SPECT, implying that the presence of linguistic abnormalities might precede the substantia nigra degeneration.Impaired linguistic function at the prodromal disease phase (RBD) might serve as a language-universal predictor for synucleinopathies. In one longitudinal study23, a narration task was submitted to 180 idiopathic RBD subjects and 149 controls speaking Czech, English, German, French and Italian for elicitation of three linguistic (content richness, vocabulary range and sentence complexity) and two acoustic features (articulatory pace and pause durations) using an automated approach. The overall phenoconversion over a 5-year follow-up on average was linked to lower content richness, slower articulation rate and prolonged pauses. Among those who developed a certain synucleinopathy, the conversion of dementia and PD with MCI compared to PD without MCI was associated with greater baseline severity of linguistic impairment (odds ratio (OR) = 109.18). Specifically, lower content richness was linked to a higher risk of PD with MCI (OR = 7.46) and restricted vocabulary range was associated with higher risks of dementia (OR = 4.08) compared to PD alone. The across-languages applicability of language as well as acoustic markers is also reflected in dysarthria treatments. For instance, following Lee Silverman Voice Treatment, vocal intensity was increased in Cantonese, Japanese, French, Mandarin and Italian speakers with PD, while intonation was improved in Portuguese, German and Cantonese speakers80. Nonetheless, other features, such as VSA and rhythm, appeared to be language-specific, and further investigations are required.Syntactic deficits might even be a preclinical sign of PD. In a small cohort comprising 33 people with sporadic PD without risk mutations, 8 with genetic PD (PARK2 or LRRK2 mutations), 9 asymptomatic first-degree relatives of people with genetic PD and 56 healthy controls with no familial history of PD81, group comparisons revealed that both sporadic and genetic PD groups performed worse on all executive and language tasks (e.g., action semantics, object semantics and action naming) compared to controls. Of interest, asymptomatic mutation carriers displayed significantly poorer syntactic comprehension than controls, implicating the role PARK2 or LRRK2 mutations may have on language-processing mechanisms. Nonetheless, this preliminary evidence comes with caveats due to the extremely limited sample sizes, and replication in larger datasets is needed.Other studies focused on early-stage PD have reported faster speaking rates characterised by short rapid output segments but longer hesitations and less word production55, impairments in alternating semantic verbal fluency (e.g., naming animals and then furniture and then alternating back and forth)82 and pragmatic skills (the ability to adapt communication techniques in terms of language comprehension and production to different social contexts)83. However, whether these symptoms are manifested in the prodromal phases is uncertain and needs to be explored in future research.Speech and language markers for differential diagnosis from other neurodegenerative conditionsPD and atypical Parkinson’s syndromes such as MSA and progressive supranuclear palsy (PSP) present with overlapping clinical features84, making accurate diagnosis challenging, especially on initial clinical evaluation. However, a few studies have highlighted that the distinct pathophysiology of atypical Parkinsonism results in unique speech symptom profiles, and detailed acoustic and language evaluation can help discriminate PD from MSA and PSP33,85,86, even in the early disease process34,35,87.In comparison to PD, speech impairment in atypical Parkinsonism is generally more pronounced34,35,85,86,87,88, possibly related to its more widespread and severe pathology and more rapid disease progression84. For instance, greater severity of imprecise consonants33,85,86, articulatory decay33,86 and diadochokinetic irregularity33,86, slower diadochokinetic and speech rate33,86 and longer pause times34,86 are frequently observed in MSA and PSP compared to PD.PD dysarthria presents mainly as hypokinetic35,88, while dysarthria in atypical Parkinsonism, in particular MSA and PSP, is often mixed involving differing combinations of hypokinetic, spastic and ataxic components86,87,88. More specifically, the speech of people with MSA is predominantly hypokinetic-ataxic, characterised by strained-strangled voice quality87,88, excessive pitch and loudness fluctuations87,88, vocal tremor87,88, variable rate of speech33 and imprecise consonants85; whereas speech in PSP is described as hypokinetic-spastic with strained-strangled voice quality87, varied speech rate33,88, stuttering like behaviour and involuntary syllable or word repetition89.Recent work suggests that diverging speech patterns may allow for better separation of PD, MSA and PSP groups. For instance, maximum phonation time, ataxia symptoms (tremulous voice, uncontrolled pitch and loudness variations during sustained phonation) and spastic features (strained, squeezing voice, phonation breaks and voiceless ratio during the same task) can split MSA and PD87. Spastic speech dimensions can also differentiate people with PSP and PD87. When using a combination of these distinct speech patterns, PD can be accurately distinguished from MSA and PSP with an accuracy of 87–95%33,88, even at early disease stages34.Atypical Parkinsonism might affect language to a greater extent compared to PD, though evidence is equivocal. When matching for age, education and global cognitive function, people with PSP exhibit severe impairments in phonological and semantic fluency, reflecting greater executive function damage, followed by those with MSA and then PwPD90. The PSP and MSA groups also present with more severe behavioural symptoms, including apathy and depression compared to PwPD90. PD and MSA groups appear to be more closely aligned on language disturbances, with the latter showing worse executive function and semantic fluency91 but similar performance in other language testing such as repetition of words and non-words, repetition of sentences, word reading, semantic association, number of phonological errors/number of words and picture description92. A combined measure encompassing the full speech and language assay might enhance group differentiation.Studies comparing PD and other neurological conditions involving movement disorders lend further support for the specificity of speech markers. PwPD and people with Huntington’s disease (PwHD) can exhibit distinct phonatory and resonatory characteristics as well as language production deficits. Based upon the assessment of sustained phonation and monologue task performance, a harsh or hoarse voice was more frequently noted in PwPD compared to controls32, whereas PwHD can present with more severe dysphonia and breathiness32 and increased and intermittent hypernasality, likely reflecting the velopharyngeal mechanism’s choreatic motions93. In terms of language functions, while both groups produced fewer grammatical sentences (e.g., subject-verb-objective) and verbal content that was less informative than controls, PwHD tend to produce shorter utterances with greater syntactic simplification than PwPD, despite performing similarly on cognitive and motor speech tasks57.PD can also be accurately separated from essential tremor (ET) even across language groups with the use of automated speech signal processing and machine learning algorithms94. A recent study proposed a classification model for PD and ET based on articulatory, phonatory and prosodic indices elicited via rapid syllable repetition and monologue tasks94. Data demonstrated that the models trained in German and Spanish achieved a classification accuracy of up to 81% for monologue and 86% for syllable repetition when applied to a Czech-speaking cohort.While existing evidence underpins the promising diagnostic and screening values of objective speech assessment in differentiating PD from other Parkinsonian syndromes, it should be noted that most of these studies were limited by very small sample sizes with case subgroups ranging from 10 to 30 participants. Additionally, there is evidence that speaker sex, often not accounted for in these studies, influences speech outcomes in PD and similar neurodegenerative disorders, with males generally experiencing greater impairment32,35. During early disease stages, males with MSA can display a considerably increased f0, slower speech rate and longer pauses compared to those with PD, whereas MSA females but not males or those with PD can manifest voice tremor and loudness decay35. Similarly, although both sexes exhibit poorer voice quality in PD and HD, males with PD experience more severe breathiness in speech and those with HD show more pronounced hoarseness compared to their female counterparts32. This differential effect may stem from the physiological distinctions between sexes in terms of the structure of vocal folds and larynx and sexual hormonal levels95,96, which also lead to variations in voice changes as a result of normal aging56,96,97. Thus, future studies with larger populations from multiple centres and proper sex stratifications are needed to validate these initial findings.Differential speech and language patterns among PD subtypesIt is well recognised that PD has a wide range of clinical manifestations and prognosis98,99, indicating the presence of multiple subtypes with varying disease mechanisms. Markers of different PD phenotypes are therefore needed to better allocate patient groups and guide more appropriate and individualised clinical management. Speech, the product of highly complex motor and cognitive coordination100, is particularly sensitive to neural structure damage and could potentially aid in classifying disease subtypes.Multiple studies have documented the distribution of affected speech domains across PD phenotypes. In particular, an association has been observed between the severity of speech disturbances, especially temporal disruptions and dysrhythmicity, and axial motor symptoms in PD29,37, including gait freezing101,102,103, festination104 and instability36,105,106, implicating a shared pathomechanism between speaking and walking dysfunctions. Early evidence came from a large cohort of 800 people with early PD (mean symptom duration = 2 years) from the DATATOP clinical trial monitored for approximately 14 months101. Speech performance measured subjectively via clinical rating scales, among other motor traits, is a strong risk factor for the development of freezing of gait101. Speech and handwriting abnormalities measured within the UPDRS were also related to episodes of gait freezing, but not rigidity, bradykinesia or balance, in a Levodopa “on” state102. Further, the decrease in gait freezing frequency in response to Levodopa was strongly associated with an improvement in speech. Similar relationships were observed in an “off-medication” state, where changes in gait velocity, cadence (steps/min) and stride length were correlated with changes in speech initiation time, speech rate and the number of repetitions during sentence reading in patients with gait freezing103.The link between speech and gait is reflected in other motor assays. During diadochokinetic tasks (syllable repetition), involuntary speeding up of speech (known as oral festination) correlated strongly with festination of gait when PwPD were in the “off Levodopa” state104. This also plays out where prominent articulatory-prosodic abnormalities change in line with greater postural instability/gait difficulties (PIGD)106. Moreover, the severity of overall speech impairment in the PIGD subtype was found to be greater compared to the tremor-dominant (TD) subtype36,105, with distinct characteristics allowing for differentiation between the two PD phenotypes105. Specifically, while both TD and PIGD subtypes displayed monopitch and irregular diadochokinetic rate, consonant timing deficits, abnormal pitch breaks, articulatory decay, decreased speech rate in subsequent segments and inappropriate pauses were found only in PIGD compared to controls36,105. Additionally, the correlations between speech (e.g., prolonged VOT, diadochokinetic irregularity and prolonged pauses) and gait (e.g., slower gait velocity, decreased cadence and shorter stride length) features were observed in the PIGD but not TD subgroup36. It is also worth noting that the severity of PIGD is also associated with impaired global cognition, executive function, memory and phonemic fluency107.As indicated by other lines of work, discernible patterns of dysarthria are thought to be associated with the onset and form (sporadic versus monogenic) as well as cognitive involvement of the disease. PwPD with a late onset (age of onset at ≥ 70 years) seem to exhibit more pronounced speech impairment, whereas those with an early onset (age of onset at ≤ 50 years) have greater motor dysfunctions including impaired strength of inspiratory muscle37. When comparing subgroups of newly diagnosed PD with an early or late onset prior to treatment commencement to age- and sex-matched healthy controls (n = 24 for each subgroup), monopitch, monoloudness and articulatory decay were consistently related to PD regardless of age; however, weaker inspirations (the relative loudness of respiration to speech) were a distinct feature of early-onset PD, while decreased voice quality and imprecise consonant articulation were unique characteristics of late-onset PD37.Sporadic and genetic PD can result in mixed acoustic profiles. Consistent with previous research indicating heterogeneous neuropathology between the two forms of PD108,109, preliminary data suggest that PD associated with mutations in the leucine-rich repeat kinase 2 (LRRK2) gene may exert a distinct impact on speech compared to idiopathic PD110. In one case, voice characteristics based on sustained phonation, such as entropy (calculated following sound wavelet decomposition and quantifies the degree of randomness in a voice signal), skewness of the amplitude distribution, glottis to noise excitation ratio (signal strength relative to noise) and vocal fold excitation ratio (signal strength versus noise) combined with other acoustic parameters differentiated the two subtypes with a mean sensitivity of 95% and specificity of 90%110. Differences were also observed among asymptomatic LRRK2-mutation carriers, asymptomatic relatives without LRRK2-mutations and unrelated healthy controls, although to a lesser degree (sensitivity 75–76% and specificity 78–82%). Nonetheless, the generalisability of this evidence is constrained by the limited sample size with only 7–20 individuals in each PD subgroup. Further validation in larger cohorts is necessary to confirm these initial findings.Linguistic and speech features may indicate declined cognition in PD, even in the early stages and mild forms. Greater linguistic impairment associated with MCI can be seen through reduced content density23, a smaller lexicon and more frequent dysfluencies in both PD and the RBD prodromal phase79. Pausing patterns also appear to be a useful marker of MCI in PD4,111,112. The location (e.g., pauses before and between utterances irrespective of verb class) and duration of pauses within utterances during a picture description task were found to be associated with cognitive function in PD evaluated by the Montreal Cognitive Assessment (MoCA)111,112. PwPD with MCI present with longer pauses between and within utterances as well as before action utterances and more frequent pauses before and within utterances than those with unremarkable cognition111. The association between pauses before action but not non-action utterances and cognitive dysfunction persists even after controlling for age, sex and UPDRS motor scores. Similarly, motor speech function can also reflect cognitive impairment, with lower f0 variation, CPP113 and VAI114, as well as more pronounced temporal coordination deficits112. Notably, tasks with a higher cognitive load, such as story retelling compared to reading, may more effectively capture alterations in cognition112,115.We are reaching a point where distinct patterns of speech and linguistic features across PD phenotypes can be identified, supporting their potential as valuable tools for detecting and differentiating patient subgroups, even early in the disease process, to facilitate timely and personalised interventions. However, it is important to note that nearly all current studies focused on only one aspect of the disease spectrum, and the utility of speech markers in simultaneously identifying disease subtypes needs attention.Speech and language markers of PD stages and progressionSpeech and language functioning declines as PD progresses. Several lines of research have provided initial evidence on the value of acoustic markers across different speech subsystems as well as linguistic features in estimating the progression and severity of the disease. Early investigations assessing PwPD at early (disease duration ≤ 5 years without motor fluctuations) and late stages (disease duration ≥ 6 years) as compared to controls implicate a cooccurrence between changes in the characteristics of speech and voice dysfunctions and disease severity and duration51. Based on sustained vowels, scaling/gliding singing and monologue, people at late PD stages manifest a perceptually softer and more prominent breathy voice compared to those at early stages, with vocal tremor being a unique feature of advanced PD. The extent of vocal loudness reduction, limited variability in pitch and loudness, restricted maximum phonation frequency range, jitter and breathiness deteriorates as PD progresses. Among these, monopitch and monoloudness may differentiate the three groups. Sex-specific voice changes were also observed, with higher mean and minimum f0 in advanced stages for males and a more pronounced reduction in f0 variability and maximum f0 in females. However, this pattern can occur as a part of normal aging96, and whether it signals more widespread disease pathology is unclear.Evidence from a larger PD cohort (n = 200) utilising a 2 min conservational speech task documents a gradual deterioration of speech accompanied by corresponding changes in different speech domains17. Dysphonia, including harsh voice quality, decreased volume and impaired intonation are common and predominantly impact the speaker in the initial PD stages. As the disease and overall speech impairment become more pronounced, the prevalence and degree of articulatory (e.g., imprecise articulation due to articulatory movement undershoot) and fluency impairments (e.g., difficulties in motor initiation, inappropriate pauses, syllable repetition and rushes in speech) increase, with articulation deficits being the most prevalent and significant feature in advanced stages.Efforts to apply automated speech processing and novel machine learning methods to predict disease status and severity utilising speech samples as simple as sustained phonation116,117,118,119,120 or combined with other tasks76,121 provide corroborating evidence. Analysis of acoustic indices extracted from sustained vowels over 2 years revealed increased perturbations in sound frequency and intensity in PwPD, which predicted PD progression measured by UPDRS part III with an error rate of ~26% and part IV with an error rate of ~11%116. Higher accuracies were obtained by employing more sophisticated feature selection and classification algorithms117,118,120,121, or incorporating simple motor tasks76. For instance, the UPDRS total and motor scores over an interval of ~6 months can be precisely estimated by random forest classifiers based on phonatory acoustic measures, with a discrepancy of < 2 points from the clinician’s ratings117,118. The integration of extra motor information such as posture, gait and finger tapping alongside speech features also achieves accurate predictions for both the presence and status of the disease76. Notably, the extension to smartphone applications marks a pivotal juncture in remote symptom monitoring on a more regular basis76,117,118, facilitating more effective disease management and improved patient care.While multiple studies provide compelling evidence on the potential of speech markers in tracking disease progression, important methodological limitations are noted; most of them either adopted a cross-sectional approach51, relied solely on case observations116,117,118 or both17,119,120,121. Small sample sizes ( < 100) used in model training and lack of validation in independent populations also raise concerns about the potential overfitting.Longitudinal comparisons with control groups generally provide more power in assuring that changes in speech are the result of disease progression rather than natural aging or individual variability in symptomatology. Evidence from this study design can also provide insights into the cooccurrence of progressive voice and speech impairment with the overall disease course. An instable pace and pace acceleration during syllable during syllable repetitions may present in PD with further deterioration over the disease course, as observed over a period of 34 months on average (range = 12–88 months)28. These features evaluated by diadochokinetic tasks differentiated PwPD from healthy individuals at baseline, though data is not available for controls at follow-up examinations28. In this cohort, alterations in speech functions are independent of the UPDRS motor scores, suggesting that impaired syllable repetition could potentially be used as a marker of non-dopaminergic disease progression in PD. However, PD can manifest as varying speaking rates, and evidence regarding the characteristics and temporal changes of over time in the disease course is often mixed. Some data suggests that PwPD speak at faster rates than those without the condition as evaluated by different speech tasks (standard passage reading122 or narration of a self-chosen topic123) with subsequent increases in speech rate and impairments over time, whereas others indicate a slower speech rate31 or a gradual decline towards normal patterns124.Sex may play an important role. Data from a German-speaking cohort revealed that although the total speech rate of males with PD was significantly faster than age-matched male controls at baseline, this difference diminished after a mean interval of 25 months (range = 7–79 months) and was comparable to that of controls at follow-up124. In PD females, while there is a progressive decline in prosodic abilities indicated by notably reduced lower range and standard deviation of f0 compared to female controls at both time points, fluctuations in speech rate are not observed124. Altered cognition in PD can also have an impact. Slowed speech rate, which further declined over time, was observed in PD complicated by dementia compared to non-demented PD31, possibly related to the consequent difficulties in planning, initiating and maintaining speech output.Changes in speech loudness and output length also vary based on the specific cohort sampled in a study. Over an average timeframe of 3.7 years, vocal loudness assessed within a reading task increased in one PD cohort compared to controls122 but remained stable in the other cohort using a narration task123. Similarly, utterance length decreased from the first to second time points in one PD cohort to a greater extent than in controls123, whereas no significant changes were observed either between groups or between time points in another cohort122. Discrepancies between studies might be attributed to 1) small sample sizes, ranging from only eight to fifty individuals with and without PD, respectively, which could substantially impede the reliability and generalisability of the findings; 2) bias introduced by the heterogeneity among PwPD in terms of disease duration upon enrolment, treatment or medication regimes and possibly disease subtypes; 3) variable time intervals between data collection points among PwPD and between PD and control groups, although speech performance was found to be independent of the intervals between examinations124 or 4) intrinsic variability in each person’s pathological processes and disease course.Imprecise vowel articulation could serve as a marker for disease status as its deterioration likely parallels the progression of the disease and especially axial motor symptoms29. Using the triangular VSA (tVSA) and VAI of vowels /a/, /i/ and /u/ extracted from a standardised passage reading task, PwPD exhibit progressively impaired vowel articulation over an interval of ~33 months, with VAI appearing to be more sensitive to altered vowels and further decline in performance over time29. Of interest, although being independent of overall motor impairment evaluated by UPDRS part III or disease duration, a decline in both tVSA and VAI correlated with axial gait dysfunctions as measured by the UPDRS gait subscores.Alterations in other speech modalities may also reflect disease progress. Assessment on a broader scale, encompassing both perceptually and acoustically measured phonation, articulation, prosody and fluency along with perceptual intelligibility, in a larger sample (80 PwPD and 60 age- and sex-matched controls) revealed that impaired voice and speech functions may be present even in the mild form of the disease and continue to decline in the disease course27. Based on the performance in reading and sustained vowel tasks, PwPD showed a decline in all perceptually evaluated speech modalities and acoustic parameters including shimmer, noise-to-harmonics ratio, net speech rate, VAI and stop consonant articulation over 12 to 88 months27. Notably, perceptual speech decline correlated with the baseline Hoehn-Yahr scale disease stages and UPDRS motor scores, while pause ratio and percentage of pauses within polysyllabic words from acoustic analysis correlated with disease stages at baseline and follow-up.Longitudinal research, though relatively limited, suggests that the decline in higher-level language functioning and speech-related breathing also occurs in tandem with the progression of PD. Compensatory speech respiratory adaptations (e.g., increase in lung volume excursion and vital capacity per syllable to maintain subglottal pressure for speech production) that are typically seen in normal aging may be compromised in PD, which can instead induce opposite alterations in these mechanisms such as lowered lung volume at initiation and termination of speech122 and decreased lung volume excursion123. PwPD, in comparison to healthy individuals, also experience reduced oral fluency and inappropriate pauses that worsen over disease progression54. Less frequent breath pauses at major syntactic boundaries and periods and more breath pauses at non-punctuational locations were also noted as PD progressed. In particular, breath pauses at non-syntactic boundaries and linguistic errors including repetitions of multiple words, additions during passage reading and other deviations from the given content were identified as the main attributing factors to overall speech impairment54.Language production capabilities may also progressively decline over time, especially in the presence of altered cognitive capacity. When evaluated on a picture narration task, the verbal output is shorter and less informative in PwPD with dementia and people with DLB compared to healthy individuals and PwPD without dementia31. PwPD with dementia also experienced a significant decline in all language measures (e.g., fluency, grammar and content density) during a mean period of 38 months, while PwPD without dementia remained stable. Speech and language impairments, according to the structural MRI scans, were related to grey matter atrophy in regions important for language performance but not motor brain regions. However, this evidence is still tentative considering the constrained sample size.Taken together, initial explorations in small cohorts have shed light on voice, speech and language changes occurring over the course of PD that are independent of normal aging. Acoustic and language features appear sensitive for monitoring disease status as they alter progressively over time even though motor symptoms remain widely stable.Gaps and future directionsCurrently available evidence seems to confirm that speech features hold promise for early disease detection, prediction as well as disease progress tracking. Validated speech and language analysis tools can become a viable component of healthcare practice during health check-ups and screening for at-risk individuals, such as older adults and people with a family history of PD or early motor symptoms.However, findings are somewhat conflicting especially regarding the longitudinal changes of speech and language functioning, likely due to heterogeneity among the underlying study populations. Replication in larger cohorts across ancestral and language groups (e.g., non-European populations) and studies following people newly diagnosed with PD before the commencement of treatment are necessary to substantiate existing findings and to elucidate the natural progression of speech-related symptoms.Notably, existing research has primarily anchored on motor speech features, particularly phonation, articulation and prosody, while largely overlooking cognitive-linguistic abnormalities. To gain a comprehensive understanding of how PD impacts speech, it is essential to explore a broader spectrum of related impairments. This includes voice quality, respiration and resonance, as well as linguistic and language aspects such as lexicon, grammar, syntax, verbal output content and fluency. Integrating both motor and cognitive-linguistic metrics will help delineate a complete picture of speech and language changes in PD and identify the best combination of markers for effective disease detection and prediction.In a similar vein, while distinct speech patterns have been linked to different disease phenotypes, studies often concentrate on isolated subtypes. However, evaluation of speech features across multiple PD subtypes can reveal overlapping and unique patterns that might not be apparent when studied in isolation, which can improve diagnostic accuracy and patient group allocation. Thus, further research on clinical phenotyping through a holistic lens is needed to determine the utility of speech indicators for concurrent recognition of various PD subtypes.Future studies exploring functional communication outside of structured, clinical settings are also necessary to assess how effectively speech-related markers perform in practical, everyday situations. This insight is crucial for developing tools that enable frequent and remote monitoring, thereby facilitating ongoing management and treatment of PD.Integration of additional factors alongside speech and language features has not yet been thoroughly investigated but could potentially enhance the prediction, diagnosis and monitoring of PD. For instance, incorporating extra information on early motor and non-motor symptoms, as well as demographics, lifestyle and environmental risk factors, many of which are often readily available or easily measured, could further refine the diagnostic process. Combining these diverse elements can help achieve a more accurate assessment of PD, leading to better-informed diagnostic and management strategies.Finally, future research is required to uncover the pathogenesis of speech and language deficits and their associated risk profiles. While some aspects of the pathophysiological mechanisms underlying these symptoms in PD have been explored, a comprehensive understanding remains elusive. The enhanced knowledge could provide valuable insights into a more nuanced understanding of the overall disease onset and progression, facilitate the identification of disease-modifying targets and improve early intervention strategies.In summary, PD is a complex neurodegenerative disorder that poses a significant global health burden. Diagnosis currently hinges solely on the presence of its key motor symptoms, which reflects considerable neuronal damage and restricts the effectiveness of neuroprotective treatments. Early detection of PD could be improved with the integration of speech-related features, which often appear early and progress with the disease, allowing for timely intervention. Specific patterns of speech and language impairment across different PD phenotypes could also aid in better patient classification and inform personalised clinical management strategies. Moreover, the simplicity and low cost of speech analysis, combined with automated processing on digital platforms, enable remote and frequent evaluations accessible to a wider population. However, further research with larger samples and a holistic approach is required to validate these preliminary findings before clinical application.
Data availability
This narrative review does not involve original data collection. All data supporting the conclusions of this review are available in the reference list and can be accessed through the respective journals.
ReferencesPoewe, W. et al. Parkinson disease. Nat. Rev. Dis. Prim. 3, 17013 (2017).PubMed
Google Scholar
Statistics Parkinson’s Australia [updated 06 May 2024; cited 2024 20 May]. Available from: https://www.parkinsons.org.au/statistics/#:~:text=Young%20Onset%20Parkinson’s%20makes%20up,to%20the%20World%20Health%20Organisation.Magee, M., Copland, D. & Vogel, A. P. Motor speech and non-motor language endophenotypes of Parkinson’s disease. Expert Rev. Neurother. 19, 1191–1200 (2019).CAS
PubMed
Google Scholar
Smith, K. M. & Caplan, D. N. Communication impairment in Parkinson’s disease: Impact of motor and cognitive symptoms on speech and language. Brain Lang. 185, 38–46 (2018).PubMed
Google Scholar
Hoglinger, G. U. et al. A biological classification of Parkinson’s disease: the SynNeurGe research diagnostic criteria. Lancet Neurol. 23, 191–204 (2024).PubMed
Google Scholar
Cardoso, F. et al. A Statement of the MDS on Biological Definition, Staging, and Classification of Parkinson’s Disease. Mov. Disord. 39, 259–266 (2024).PubMed
Google Scholar
Cheng, H. C., Ulane, C. M. & Burke, R. E. Clinical progression in Parkinson disease and the neurobiology of axons. Ann. Neurol. 67, 715–725 (2010).PubMed
PubMed Central
Google Scholar
Greffard, S. et al. Motor Score of the Unified Parkinson Disease Rating Scale as a Good Predictor of Lewy Body–Associated Neuronal Loss in the Substantia Nigra. Arch. Neurol. 63, 584–588 (2006).PubMed
Google Scholar
Sarkar, S., Raymick, J. & Imam, S. Neuroprotective and Therapeutic Strategies against Parkinson’s Disease: Recent Perspectives. Int J. Mol. Sci. 17, 904 (2016).PubMed
PubMed Central
Google Scholar
Athauda, D. & Foltynie, T. The ongoing pursuit of neuroprotective therapies in Parkinson disease. Nat. Rev. Neurol. 11, 25–40 (2015).CAS
PubMed
Google Scholar
Gouda, N. A., Elkamhawy, A. & Cho, J. Emerging Therapeutic Strategies for Parkinson’s Disease and Future Prospects: A 2021 Update. Biomedicines 10, 371 (2022).CAS
PubMed
PubMed Central
Google Scholar
Högl, B., Stefani, A. & Videnovic, A. Idiopathic REM sleep behaviour disorder and neurodegeneration — an update. Nat. Rev. Neurol. 14, 40–55 (2018).PubMed
Google Scholar
Miglis, M. G. et al. Biomarkers of conversion to α-synucleinopathy in isolated rapid-eye-movement sleep behaviour disorder. Lancet Neurol. 20, 671–684 (2021).CAS
PubMed
PubMed Central
Google Scholar
Zhang, J., Xu, C. Y. & Liu, J. Meta-analysis on the prevalence of REM sleep behavior disorder symptoms in Parkinson’s disease. BMC Neurol. 17, 23 (2017).PubMed
PubMed Central
Google Scholar
Zhang, X., Sun, X., Wang, J., Tang, L. & Xie, A. Prevalence of rapid eye movement sleep behavior disorder (RBD) in Parkinson’s disease: a meta and meta-regression analysis. Neurol. Sci. 38, 163–170 (2017).PubMed
Google Scholar
Sateia, M. J. International classification of sleep disorders. Chest 146, 1387–1394 (2014).PubMed
Google Scholar
Ho, A. K., Iansek, R., Marigliani, C., Bradshaw, J. L. & Gates, S. Speech Impairment in a Large Sample of Patients with Parkinson’s Disease. Behav. Neurol. 11, 327643 (1999).
Google Scholar
Logemann, J. A., Fisher, H. B., Boshes, B. & Blonsky, E. R. Frequency and cooccurrence of vocal tract dysfunctions in the speech of a large sample of Parkinson patients. J. Speech Hear Disord. 43, 47–57 (1978).CAS
PubMed
Google Scholar
Ma, A., Lau, K. K. & Thyagarajan, D. Voice changes in Parkinson’s disease: What are they telling us? J. Clin. Neurosci. 72, 1–7 (2020).PubMed
Google Scholar
Moreau, C. & Pinto, S. Misconceptions about speech impairment in Parkinson’s disease. Mov. Disord. 34, 1471–1475 (2019).PubMed
Google Scholar
Sapir, S. Multiple factors are involved in the dysarthria associated with Parkinson’s disease: a review with implications for clinical practice and research. J. Speech Lang. Hear Res. 57, 1330–1343 (2014).PubMed
Google Scholar
Postuma, R. B., Lang, A. E., Gagnon, J. F., Pelletier, A. & Montplaisir, J. Y. How does parkinsonism start? Prodromal parkinsonism motor changes in idiopathic REM sleep behaviour disorder. Brain 135, 1860–1870 (2012).CAS
PubMed
Google Scholar
Subert, M. et al. Spoken Language Alterations can Predict Phenoconversion in Isolated Rapid Eye Movement Sleep Behavior Disorder: A Multicenter Study. Ann. Neurol. 95, 530–543 (2024).PubMed
Google Scholar
Goldman, J. G. et al. Delivering multidisciplinary rehabilitation care in Parkinson’s disease: an international consensus statement. J. Parkinson’s Dis. 14, 135–166 (2024).
Google Scholar
Rusz, J. et al. Speech Biomarkers in Rapid Eye Movement Sleep Behavior Disorder and Parkinson Disease. Ann. Neurol. 90, 62–75 (2021).PubMed
PubMed Central
Google Scholar
Jeancolas, L. et al. Voice characteristics from isolated rapid eye movement sleep behavior disorder to early Parkinson’s disease. Parkinsonism Relat. Disord. 95, 86–91 (2022).PubMed
Google Scholar
Skodda, S., Groenheit, W., Mancinelli, N. & Schlegel, U. Progression of Voice and Speech Impairment in the Course of Parkinson’s Disease: A Longitudinal Study. Parkinsons Dis.-Us 2013, 389195 (2013).CAS
Google Scholar
Skodda, S., Flasskamp, A. & Schlegel, U. Instability of Syllable Repetition as a Marker of Disease Progression in Parkinson’s Disease: A Longitudinal Study. Mov. Disord. 26, 59–64 (2011).PubMed
Google Scholar
Skodda, S., Grönheit, W. & Schlegel, U. Impairment of Vowel Articulation as a Possible Marker of Disease Progression in Parkinson’s Disease. Plos One 7, e32132 (2012).CAS
PubMed
PubMed Central
Google Scholar
Harel, B. T., Cannizzaro, M. S., Cohen, H., Reilly, N. & Snyder, P. J. Acoustic characteristics of Parkinsonian speech: a potential biomarker of early disease progression and treatment. J. Neurolinguist. 17, 439–453 (2004).
Google Scholar
Ash, S. et al. Longitudinal decline in speech production in Parkinson’s disease spectrum disorders. Brain Lang. 171, 42–51 (2017).PubMed
PubMed Central
Google Scholar
Simek M., Kouba T., Novotny M., Tykalova T., Rusz J. Comparison of acoustic measures of dysphonia in Parkinson’s disease and Huntington’s disease: Effect of sex and speaking task.Daoudi, K., Das, B., Tykalova, T., Klempir, J. & Rusz, J. Speech acoustic indices for differential diagnosis between Parkinson’s disease, multiple system atrophy and progressive supranuclear palsy. NPJ Parkinsons Dis. 8, 142 (2022).PubMed
PubMed Central
Google Scholar
Kang, K. et al. Utilizing speech analysis to differentiate progressive supranuclear palsy from Parkinson’s disease. Parkinsonism Relat. Disord. 115, 105835 (2023).PubMed
PubMed Central
Google Scholar
Huh, Y. E. et al. Differences in early speech patterns between Parkinson variant of multiple system atrophy and Parkinson’s disease. Brain Lang. 147, 14–20 (2015).PubMed
Google Scholar
Rusz, J. et al. Speech and gait abnormalities in motor subtypes of de-novo Parkinson’s disease. CNS Neurosci. Ther. 29, 2101–2110 (2023).PubMed
PubMed Central
Google Scholar
Rusz, J., Tykalová, T., Novotný, M., Růžička, E. & Dušek, P. Distinct patterns of speech disorder in early-onset and late-onset de-novo Parkinson’s disease. NPJ Parkinsons Dis. 7, 98 (2021).PubMed
PubMed Central
Google Scholar
Vogel, A. P. et al. Quantitative Speech Assessment in Ataxia-Consensus Recommendations by the Ataxia Global Initiative Working Group on Digital-Motor Markers. Cerebellum 23, 1128–1134 (2024).PubMed
Google Scholar
Rusz, J., Tykalova, T., Ramig, L. O. & Tripoliti, E. Guidelines for Speech Recording and Acoustic Analyses in Dysarthrias of Movement Disorders. Mov. Disord. 36, 803–814 (2021).PubMed
Google Scholar
Noffs, G. et al. Speech metrics, general disability, brain imaging and quality of life in multiple sclerosis. Eur. J. Neurol. 28, 259–268 (2021).CAS
PubMed
Google Scholar
Boschi, V. et al. Connected Speech in Neurodegenerative Language Disorders: A Review. Front Psychol. 8, 269 (2017).PubMed
PubMed Central
Google Scholar
Shao, Z., Janse, E., Visser, K. & Meyer, A. S. What do verbal fluency tasks measure? Predictors of verbal fluency performance in older adults. Front. Psychol. 5, 772 (2014).PubMed
PubMed Central
Google Scholar
Henry, J. D. & Crawford, J. R. Verbal fluency deficits in Parkinson’s disease: A meta-analysis. J. Int Neuropsych. Soc. 10, 608–622 (2004).
Google Scholar
Sussman, J. E. & Tjaden, K. Perceptual measures of speech from individuals with Parkinson’s disease and multiple sclerosis: intelligibility and beyond. J. Speech Lang. Hear Res. 55, 1208–1219 (2012).PubMed
Google Scholar
Jacobson, B. H. et al. The Voice Handicap Index (VHI). Am. J. Speech-Lang. Pathol. 6, 66–70 (1997).
Google Scholar
Donovan, N. J., Kendall, D. L., Young, M. E. & Rosenbek, J. C. The Communicative Effectiveness Survey: Preliminary Evidence of Construct Validity. Am. J. Speech-Lang. Pathol. 17, 335–347 (2008).PubMed
Google Scholar
Hogikyan, N. D. The Voice-Related Quality of Life (V-RQOL) Measure: History and Ongoing Utility of a Validated Voice Outcomes Instrument. Perspect. Voice Voice Disord. 14, 3–5 (2004).
Google Scholar
Walshe, M., Peach, R. K. & Miller, N. Dysarthria Impact Profile: development of a scale to measure psychosocial effects. Int. J. Lang. Commun. Disord. 44, 693–715 (2009).PubMed
Google Scholar
Atkinson-Clement, C. et al. Psychosocial Impact of Dysarthria: The Patient-Reported Outcome as Part of the Clinical Management. Neurodegenerative Dis. 19, 12–21 (2019).
Google Scholar
Vogel A., et al. Development and validation of the dysarthria impact scale. International Congress for Ataxia Research; Dallas; (2022).Holmes, R. J., Oates, J. M., Phyland, D. J. & Hughes, A. J. Voice characteristics in the progression of Parkinson’s disease. Int J. Lang. Commun. Disord. 35, 407–418 (2000).CAS
PubMed
Google Scholar
Rusz, J. et al. Imprecise vowel articulation as a potential early marker of Parkinson’s disease: Effect of speaking task. J. Acoust. Soc. Am. 134, 2171–2181 (2013).PubMed
Google Scholar
Darley, F. L., Aronson, A. E. & Brown, J. R. Differential Diagnostic Patterns of Dysarthria. J. Speech Hearing Res. 12, 246–269 (1969).CAS
PubMed
Google Scholar
Darling-White, M. & Huber, J. E. The Impact of Parkinson’s Disease on Breath Pauses and Their Relationship to Speech Impairment: A Longitudinal Study. Am. J. Speech-Lang. Pat. 29, 1910–1922 (2020).
Google Scholar
Illes, J. Neurolinguistic features of spontaneous language production dissociate three forms of neurodegenerative disease: Alzheimer’s, Huntington’s, and Parkinson’s. Brain Lang. 37, 628–642 (1989).CAS
PubMed
Google Scholar
Tykalova, T. et al. Effect of Ageing on Acoustic Characteristics of Voice Pitch and Formants in Czech Vowels. J. Voice 35, 931 e21–e33 (2021).PubMed
Google Scholar
Murray, L. L. Spoken Language Production in Huntington’s and Parkinson’s Diseases. J. Speech, Lang., Hearing Res. 43, 1350–1366 (2000).CAS
Google Scholar
García, A. M. et al. How language flows when movements don’t: An automated analysis of spontaneous discourse in Parkinson’s disease. Brain Lang. 162, 19–28 (2016).PubMed
Google Scholar
Illes, J., Metter, E. J., Hanson, W. R. & Iritani, S. Language production in Parkinson’s disease: Acoustic and linguistic considerations. Brain Lang. 33, 146–160 (1988).CAS
PubMed
Google Scholar
Rohl, A. et al. Speech dysfunction, cognition, and Parkinson’s disease. Prog. Brain Res 269, 153–173 (2022).PubMed
PubMed Central
Google Scholar
Ariatti, A., Benuzzi, F. & Nichelli, P. Recognition of emotions from visual and prosodic cues in Parkinson’s disease. Neurol. Sci. 29, 219–227 (2008).PubMed
Google Scholar
Mattavelli, G. et al. Facial expressions recognition and discrimination in Parkinson’s disease. J. Neuropsychol. 15, 46–68 (2021).PubMed
Google Scholar
Ricciardi, L. et al. Facial Emotion Recognition and Expression in Parkinson’s Disease: An Emotional Mirror Mechanism? PLoS One 12, e0169110 (2017).PubMed
PubMed Central
Google Scholar
Braak, H., Ghebremedhin, E., Rüb, U., Bratzke, H. & Del Tredici, K. Stages in the development of Parkinson’s disease-related pathology. Cell Tissue Res. 318, 121–134 (2004).PubMed
Google Scholar
Gómez-Vilda, P. et al. Parkinson Disease Detection from Speech Articulation Neuromechanics. Front. Neuroinformatics 11, 56 (2017).
Google Scholar
Selby G. Disease of the basal ganglia. Handbook of clinical neurology. 173–211. (1968).Rusz, J. et al. Dysprosody in Isolated REM Sleep Behavior Disorder with Impaired Olfaction but Intact Nigrostriatal Pathway. Mov. Disord. 37, 619–623 (2022).CAS
PubMed
Google Scholar
Caverlé, M. W. J. & Vogel, A. P. Stability, reliability, and sensitivity of acoustic measures of vowel space: A comparison of vowel space area, formant centralization ratio, and vowel articulation index. J. Acoustical Soc. Am. 148, 1436–1444 (2020).
Google Scholar
Skrabal, D. et al. Articulatory undershoot of vowels in isolated REM sleep behavior disorder and early Parkinson’s disease. NPJ Parkinsons Dis. 8, 137 (2022).PubMed
PubMed Central
Google Scholar
Singh, S. & Xu, W. Y. Robust Detection of Parkinson’s Disease Using Harvested Smartphone Voice Data: A Telemedicine Approach. Telemed. E-Health 26, 327–334 (2020).
Google Scholar
Agarwal A., Chandrayan S., Sahu S. S., editors. Prediction of Parkinson’s disease using speech signal with Extreme Learning Machine. 2016 International Conference on Electrical, Electronics, and Optimization Techniques (ICEEOT); 2016 3–5 March (2016).Tsanas, A., Little, M. A., McSharry, P. E., Spielman, J. & Ramig, L. O. Novel Speech Signal Processing Algorithms for High-Accuracy Classification of Parkinson’s Disease. Ieee T Bio-Med Eng. 59, 1264–1271 (2012).
Google Scholar
Hlavnicka, J. et al. Automated analysis of connected speech reveals early biomarkers of Parkinson’s disease in patients with rapid eye movement sleep behaviour disorder. Sci. Rep. 7, 12 (2017).PubMed
PubMed Central
Google Scholar
Tracy, J. M., Özkanca, Y., Atkins, D. C. & Ghomi, R. H. Investigating voice as a biomarker: Deep phenotyping methods for early detection of Parkinson’s disease. J. Biomed. Inf. 104, 103362 (2020).
Google Scholar
Illner, V. et al. Smartphone Voice Calls Provide Early Biomarkers of Parkinsonism in Rapid Eye Movement Sleep Behavior Disorder. Mov. Disord. 39, 1752–1762 (2024).CAS
PubMed
Google Scholar
Arora, S. et al. Detecting and monitoring the symptoms of Parkinson’s disease using smartphones: A pilot study. Parkinsonism Relat. Disord. 21, 650–653 (2015).CAS
PubMed
Google Scholar
Kouba, T., Illner, V. & Rusz, J. Study protocol for using a smartphone application to investigate speech biomarkers of Parkinson’s disease and other synucleinopathies: SMARTSPEECH. BMJ Open 12, e059871 (2022).PubMed
PubMed Central
Google Scholar
Arora, S. & Tsanas, A. Assessing Parkinson’s Disease at Scale Using Telephone-Recorded Speech: Insights from the Parkinson’s Voice Initiative. Diagnostics 11, 1892 (2021).PubMed
PubMed Central
Google Scholar
Subert, M. et al. Linguistic Abnormalities in Isolated Rapid Eye Movement Sleep Behavior Disorder. Mov. Disord. 37, 1872–1882 (2022).CAS
PubMed
Google Scholar
Levy, E. S. & Moya-Galé, G. Revisiting Dysarthria Treatment Across Languages: The Hybrid Approach. J. Speech, Lang., Hearing Res. 67, 2893–2902 (2024).
Google Scholar
García, A. M. et al. Language Deficits as a Preclinical Window into Parkinson’s Disease: Evidence from Asymptomatic Parkin and Dardarin Mutation Carriers. J. Int Neuropsych Soc. 23, 150–158 (2017).
Google Scholar
Ellfolk, U. et al. Striatal volume is related to phonemic verbal fluency but not to semantic or alternating verbal fluency in early Parkinson’s disease. J. Neural Transm. (Vienna) 121, 33–40 (2014).PubMed
Google Scholar
Baraldi, M. A. et al. Pragmatic abilities in early Parkinson’s disease. Brain Cogn. 150, 105706 (2021).PubMed
Google Scholar
Deutschländer, A. B., Ross, O. A., Dickson, D. W. & Wszolek, Z. K. Atypical parkinsonian syndromes: a general neurologist’s perspective. Eur. J. Neurol. 25, 41–58 (2018).PubMed
Google Scholar
Tykalova, T., Rusz, J., Klempir, J., Cmejla, R. & Ruzicka, E. Distinct patterns of imprecise consonant articulation among Parkinson’s disease, progressive supranuclear palsy and multiple system atrophy. Brain Lang. 165, 1–9 (2017).PubMed
Google Scholar
Rusz, J. et al. Distinctive speech signature in cerebellar and parkinsonian subtypes of multiple system atrophy. J. Neurol. 266, 1394–1404 (2019).PubMed
Google Scholar
Kowalska-Taczanowska, R., Friedman, A. & Koziorowski, D. Parkinson’s disease or atypical parkinsonism? The importance of acoustic voice analysis in differential diagnosis of speech disorders. Brain Behav. 10, e01700 (2020).PubMed
PubMed Central
Google Scholar
Rusz, J. et al. Speech disorders reflect differing pathophysiology in Parkinson’s disease, progressive supranuclear palsy and multiple system atrophy. J. Neurol. 262, 992–1001 (2015).CAS
PubMed
Google Scholar
Kluin, K. J., Foster, N. L., Berent, S. & Gilman, S. Perceptual analysis of speech disorders in progressive supranuclear palsy. Neurology 43, 563–566 (1993).CAS
PubMed
Google Scholar
Santangelo, G. et al. Comparative cognitive and neuropsychiatric profiles between Parkinson’s disease, multiple system atrophy and progressive supranuclear palsy. J. Neurol. 265, 2602–2613 (2018).PubMed
Google Scholar
Kao, A. W. et al. Cognitive and neuropsychiatric profile of the synucleinopathies: Parkinson disease, dementia with Lewy bodies, and multiple system atrophy. Alzheimer Dis. Assoc. Disord. 23, 365–370 (2009).PubMed
PubMed Central
Google Scholar
Cuoco, S. et al. The language profile in multiple system atrophy: an exploratory study. J. Neural Transm. (Vienna) 128, 1195–1203 (2021).PubMed
Google Scholar
Novotny, M. et al. Hypernasality associated with basal ganglia dysfunction: evidence from Parkinsonas disease and Huntingtonas disease. PeerJ 4, e2530 (2016).PubMed
PubMed Central
Google Scholar
Rios-Urrego, C. D., Rusz, J. & Orozco-Arroyave, J. R. Automatic speech-based assessment to discriminate Parkinson’s disease from essential tremor with a cross-language approach. npj Digital Med. 7, 37 (2024).
Google Scholar
Zamponi, V., Mazzilli, R., Mazzilli, F. & Fantini, M. Effect of sex hormones on human voice physiology: from childhood to senescence. Hormones (Athens) 20, 691–696 (2021).PubMed
Google Scholar
Rojas, S., Kefalianos, E. & Vogel, A. How Does Our Voice Change as We Age? A Systematic Review and Meta-Analysis of Acoustic and Perceptual Voice Data From Healthy Adults Over 50 Years of Age. J. Speech Lang. Hear Res. 63, 533–551 (2020).PubMed
Google Scholar
Schultz, B. G., Rojas, S., St John, M., Kefalianos, E. & Vogel, A. P. A Cross-sectional Study of Perceptual and Acoustic Voice Characteristics in Healthy Aging. J. Voice 37, 969.e23–e41 (2023).PubMed
Google Scholar
Berg, D. et al. Time to redefine PD? Introductory statement of the MDS Task Force on the definition of Parkinson’s disease. Mov. Disord. 29, 454–462 (2014).PubMed
PubMed Central
Google Scholar
Fereshtehnejad, S. M. et al. New Clinical Subtypes of Parkinson Disease and Their Longitudinal Progression: A Prospective Cohort Comparison With Other Phenotypes. JAMA Neurol. 72, 863–873 (2015).PubMed
Google Scholar
Kearney, E. & Guenther, F. H. Articulating: The Neural Mechanisms of Speech Production. Lang. Cogn. Neurosci. 34, 1214–1229 (2019).PubMed
PubMed Central
Google Scholar
Giladi, N. et al. Freezing of gait in PD. Neurology 56, 1712–1721 (2001).CAS
PubMed
Google Scholar
Bartels, A. L. et al. Relationship between freezing of gait (FOG) and other features of Parkinson’s: FOG is not correlated with bradykinesia. J. Clin. Neurosci. 10, 584–588 (2003).CAS
PubMed
Google Scholar
Park, H. K. et al. Gait freezing and speech disturbance in Parkinson’s disease. Neurological Sci. 35, 357–363 (2014).CAS
Google Scholar
Moreau, C. et al. Oral festination in Parkinson’s disease: Biomechanical analysis and correlation with festination and freezing of gait. Mov. Disord. 22, 1503–1506 (2007).PubMed
Google Scholar
Tykalová, T. et al. Speech disorder and vocal tremor in postural instability/gait difficulty and tremor dominant subtypes of Parkinson’s disease. J. Neural Transm. 127, 1295–1304 (2020).PubMed
Google Scholar
Rusz, J. et al. Defining Speech Subtypes in De Novo Parkinson Disease: Response to Long-term Levodopa Therapy. Neurology 97, e2124–e2135 (2021).CAS
PubMed
Google Scholar
Kelly, V. E. et al. Association of cognitive domains with postural instability/gait disturbance in Parkinson’s disease. Parkinsonism Relat. Disord. 21, 692–697 (2015).CAS
PubMed
PubMed Central
Google Scholar
Kalia, L. V. et al. Clinical correlations with Lewy body pathology in LRRK2-related Parkinson disease. JAMA Neurol. 72, 100–105 (2015).PubMed
PubMed Central
Google Scholar
Rocha, E. M., Keeney, M. T., Di Maio, R., De Miranda, B. R. & Greenamyre, J. T. LRRK2 and idiopathic Parkinson’s disease. Trends Neurosci. 45, 224–236 (2022).CAS
PubMed
PubMed Central
Google Scholar
Arora, S. et al. Investigating voice as a biomarker for leucine-rich repeat kinase 2-associated Parkinson’s disease. J. Parkinson’s Dis. 8, 503–510 (2018).CAS
Google Scholar
Andrade, E. I. N., Manxhari, C. & Smith, K. M. Pausing before verb production is associated with mild cognitive impairment in Parkinson’s disease. Front Hum. Neurosci. 17, 1102024 (2023).PubMed
PubMed Central
Google Scholar
Rusz, J. & Tykalová, T. Does Cognitive Impairment Influence Motor Speech Performance in De Novo Parkinson’s Disease? Mov. Disord. 36, 2980–2982 (2021).PubMed
Google Scholar
Smith, K. M., Demers-Peel, M., Manxhari, C. & Stepp, C. E. Voice Acoustic Instability During Spontaneous Speech in Parkinson’s Disease. J. Voice https://doi.org/10.1016/j.jvoice.2023.06.004 (2023).DeSilva, G. S., Upadhyay, P., Manxhari, M., Gopal, D. & Smith, K. M. Variability in Vowel Space in Parkinson’s Disease: Associations With Cognitive and Motor Impairment. J. Speech Lang. Hear Res. 67, 3566–3578 (2024).PubMed
Google Scholar
Vogel, A. P., Fletcher, J. & Maruff, P. The impact of task automaticity on speech in noise. Speech Commun. 65, 1–8 (2014).
Google Scholar
Galaz, Z. et al. Changes in Phonation and Their Relations with Progress of Parkinsonas Disease. Appl. Sci-Basel 8, 2339 (2018).
Google Scholar
Tsanas, A., Little, M. A., McSharry, P. E. & Ramig, L. O. Nonlinear speech analysis algorithms mapped to a standard metric achieve clinically useful quantification of average Parkinson’s disease symptom severity. J. R. Soc. Interface 8, 842–855 (2011).PubMed
Google Scholar
Tsanas, A., Little, M. A. & Ramig, L. O. Remote Assessment of Parkinson’s Disease Symptom Severity Using the Simulated Cellular Mobile Telephone Network. Ieee Access 9, 11024–11036 (2021).PubMed
Google Scholar
Tunc, H. C. et al. Estimation of Parkinson’s disease severity using speech features and extreme gradient boosting. Med. Biol. Eng. Comput. 58, 2757–2773 (2020).PubMed
Google Scholar
Hemmerling, D. & Wojcik-Pedziwiatr, M. Prediction and Estimation of Parkinson’s Disease Severity Based on Voice Signal. J. Voice 36, 439.e9–439.e20 (2022).PubMed
Google Scholar
Asgari, M. & Shafran, I. Predicting severity of Parkinson’s disease from speech. Annu Int Conf. IEEE Eng. Med Biol. Soc. 2010, 5201–5204 (2010).PubMed
PubMed Central
Google Scholar
Huber, J. E. & Darling-White, M. Longitudinal Changes in Speech Breathing in Older Adults with and without Parkinson’s Disease. Semin Speech Lang. 38, 200–209 (2017).PubMed
PubMed Central
Google Scholar
Darling-White, M., Anspach, Z. & Huber, J. E. Longitudinal Effects of Parkinson’s Disease on Speech Breathing During an Extemporaneous Connected Speech Task. J. Speech Lang. Hear R. 65, 1402–1415 (2022).
Google Scholar
Skodda, S., Rinsche, H. & Schlegel, U. Progression of dysprosody in Parkinson’s disease over time-a longitudinal study. Mov. Disord. 24, 716–722 (2009).PubMed
Google Scholar
Download referencesAcknowledgementsF.C. thanks scholarship support from The University of Queensland and QIMR Berghofer Medical Research Institute. M.E.R. thanks fellowship support from the Rebecca L Cooper Medical Research Foundation (F20231230). P.G. is supported by an NHMRC Investigator Grant (#1173390). A.P.V. is Chief Scientific Officer of Redenlab Inc., a speech neuroscience company.Author informationAuthors and AffiliationsBrain & Mental Health Program, QIMR Berghofer Medical Research Institute, Brisbane, QLD, AustraliaFangyuan Cao & Miguel E. RenteriaSchool of Biomedical Sciences, Faculty of Medicine, The University of Queensland, Brisbane, QLD, AustraliaFangyuan Cao, Puya Gharahkhani & Miguel E. RenteriaCentre for Neuroscience of Speech, The University of Melbourne, Victoria, AustraliaAdam P. VogelRedenlab, Melbourne, AustraliaAdam P. VogelPopulation Health Program, QIMR Berghofer Medical Research Institute, Brisbane, QLD, AustraliaPuya GharahkhaniAuthorsFangyuan CaoView author publicationsYou can also search for this author inPubMed Google ScholarAdam P. VogelView author publicationsYou can also search for this author inPubMed Google ScholarPuya GharahkhaniView author publicationsYou can also search for this author inPubMed Google ScholarMiguel E. RenteriaView author publicationsYou can also search for this author inPubMed Google ScholarContributionsF.C.: Contributed to the conceptualisation of the study, conducted the literature review and led the drafting of the manuscript. A.P.V. and M.E.R. supervised the study and reviewed and edited the manuscript. P.G. advised and reviewed the manuscript. All authors have read and approved the final version of the manuscript.Corresponding authorCorrespondence to
Miguel E. Renteria.Ethics declarations
Competing interests
The authors declare no competing interests.
Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissionsAbout this articleCite this articleCao, F., Vogel, A.P., Gharahkhani, P. et al. Speech and language biomarkers for Parkinson’s disease prediction, early diagnosis and progression.
npj Parkinsons Dis. 11, 57 (2025). https://doi.org/10.1038/s41531-025-00913-4Download citationReceived: 12 August 2024Accepted: 27 February 2025Published: 24 March 2025DOI: https://doi.org/10.1038/s41531-025-00913-4Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative