Speech, voice and language in Parkinson’s disease: changes and interventions

Reports of speech and voice changes in idiopathic Parkinson’s disease (PD) date back to the original 19th century descriptions. The nature of these alterations has been elucidated in extensive research over the past few decades and they are well known. The presence of language comprehension and expression changes and the salient problems these present to many people with PD (pwP) have, however, only been more recently highlighted. This review provides an overview of recent research concerning the nature and management of communication changes in PD. It includes current work on voice and articulation, but also focuses on cognitive–linguistic disturbances. Swallowing dysfunction and drooling are not covered, other than to mention in passing that the effort to control saliva and swallowing may secondarily impact on speaking and may influence the desire to communicate for social reasons. Writing/micrographia is outside the scope of this review, but it is noted in passing that in contrast with other neurological conditions, where written communication may replace or supplement spoken communication, this is seldom an option for pwP.

The scale of changes

Larger-scale surveys consistently suggest that approximately 70–80% of pwP experience changes in their voice that are noticeable to others and approximately 55% experience notable changes to their articulation. A study involving a community based cohort (n = 125) of pwP, predominantly between Hoehn and Yahr stages I–III, found that 96% of participants reported speech–voice changes, with 82% reporting being unhappy with how they spoke [1]. A total of 51% fell more than one standard deviation below the control mean on a speech intelligibility assessment. With 38% of people citing speaking changes among their top four concerns relating to their PD; 10% placed them first. Speech and voice changes are therefore neither rare nor trivial for pwP.

Indeed, speech or voice deterioration may be the initial sign to an individual or family that something is amiss years before a diagnosis is established [2]. Even after PD is confirmed, an absence of changes to speech and voice that are audible to the clinician should not mislead one to believe that no problem is present. Even in pwP without apparent dysarthria, acoustic and kinematic changes in speech can be detected [3–5]. Furthermore, long before frank alterations emerge, changes perceived by the pwP may have led them to develop maladaptive strategies in managing their voice, to have lost confidence in communication and motivation to speak and to have withdrawn from social situations or roles [6,7]. Insidious changes to interaction associated with altered theory of mind may also be expected to exert an influence here [8].

Typically, the likelihood of experiencing changes increases across the time course of the condition, although the relationship between speech and limb motor progression is not strong [1,7,9–12]. As discussed below, this probably reflects the cognitive–linguistic disturbances as opposed to pure motor control, nature of speech processing, and the possibility that speech–motor control is organized differently to limb motor control [13].

A description of changes

▪ Voice–respiration

Voice change is characterized by a loss of perceived loudness and is described as weak, hoarse and breathy [9,14]. Overall pitch range in habitual speech is reduced. Together, these lead to a perception of a monotonous voice with a lack of loudness and pitch variation. There may be difficulties initiating voice and vocal ‘freezing’ may cause loss-of-voice midsentence. Voice tremor may be present, but how many pwP demonstrate this and whether the tremulousness originates from vocal cord tremor or other sites in the vocal tract remains unclear [15,16].

Earlier views have associated decreased loudness and loss of intonation variation with the rigidity and stiffness in the laryngeal and ribcage muscles, which is analogous to explanations for gait and postural changes. This is at most a partial explanation. Reduced control of subglottal driving pressure from alterations to respiratory forces, and changes to the coordination of breathing and phonation also play a part [17,18]. The decisive additional factor, however, concerns a lack of adequate scaling of movement force, again, a feature that is in common with the execution of other movements, and the inability to judge one’s own speech intensity. When requested by others to speak up, pwP may insist they are already shouting. PwP are capable of producing voice, approaching as loud as matched controls when asked to say ‘ah’ as loud as possible, and certainly do in instances of paradoxical kinesia (i.e., fight and flight reactions to external danger), even in late-stage PD; but, they do not habitually talk as loud as control speakers [19]. These phenomena are associated with an apparent auditory perceptual impairment whereby pwP are unable to monitor their output effectively [20–22]. Furthermore, the problem of sustaining adequate voice loudness is combined with a perceived added sense of effort [23]. Overcoming these factors represents an important goal in therapy, and recent successful programs (e.g., Lee Silverman Voice Treatment [LSVT^®, LSVT Global Inc., AZ, USA] [24]), have as their rationale the requirement to recalibrate internal perceptions of loudness and sense of effort.

Other aspects of voice change are addressed under prosodic disturbance below.

▪ Articulation

Earlier investigations established that production of individual speech sounds becomes increasingly impaired in PD. Distinctions between different sounds become blurred [1,12], and weak or absent contacts lead to difficulty signaling distinctions between sounds/words (e.g., pea sounds like ‘fee’ or ‘he’ and do sounds like ‘zoo’ or ‘who’). Sounds in unstressed syllables are particularly prone to distortion or omission. The underlying problem though is not simply one of inability to attain absolute range or velocity; pwP are capable of attaining velocities and a range of movement that is comparable to speakers without PD [3,4,25]. Such findings support an interpretation of articulation changes in pwP that reflects impairment in the rest of the motor system; the source of pronunciation changes stems from initiating articulation with insufficient force and failure to consistently sustain a full range of movement across a whole phase. The findings also point to considerable variability in performance at an individual (as opposed to group) level, a point that is discussed later.

Debate continues over the nature of speech-rate changes in PD speech [26,27]. Speech festination, loss of tight articulatory contacts, elision of unstressed syllables and monopitch and monoloudness can lead perceptually to the impression that rate is accelerated even when objective measures demonstrate that the rate lies within normal boundaries, or is even slowed. However, there may be, in a subgroup of speakers, a tendency to accelerate over the course of an utterance, or speakers may produce short bursts of faster speech. The latter is often interpreted as a compensatory tactic to complete an utterance on insufficient breath or as an effect of realignment of speech rhythm to tremor cycles.

Articulation and intelligibility may be adequate in quiet, relaxed surroundings during general chat with one other person. Where there are competing tasks (e.g., walking and talking) or where processing capacity is also taxed through attention to language formulation or more complex thoughts, articulation and intelligibility may deteriorate [28–31]. This constitutes another reason why clinic-room performance should not be taken as representative of daily living ability.

▪ Prosody

A prosodic disturbance is central to the speech changes experienced by pwP [11,32–35]. Prosody refers to variation in pitch (intonation), amplitude (loudness), duration and rhythm in speech. A loss of responsiveness of the vocal cords in tensing and relaxing leads to a loss of pitch variation (i.e., the voice sounds monotone). Coupled with this is a tendency for the mean fundamental frequency of voice to rise in men and to lower in women (i.e., their voices sound inappropriately high and deep, respectively). Loss of effective control of driving air pressure from the lungs and poorer vocal cord adductory capability leads not just to a loss of loudness, but to a loss of the ability to contrast louder (stressed) syllables and words with unstressed syllables (i.e., giving the impression of monoloudness and lack of loudness variation). The result is an impairment of the ability to signal linguistic and emotional nuances in speech that depend on intonation and stress contrasts (e.g., ‘THAT’S my coat’ vs ‘that’s MY coat’, ‘that’s a lovely dress’ spoken sincerely vs implying it’s definitely not, ‘my shirt came out the wash yellow’ expressed as funny vs upset). Difficulties initiating speech and dysfluency while speaking also disrupt rhythm.

The expressive aspects of dysprosody have long been recognized. More recently, it has been confirmed that in pwP this may be part of a wider impairment of prosody perception and planning, affecting both language and broader emotional prosodic processes [36]. Thus, the difficulty in signaling pitch and amplitude variation is not solely a mechanical or neuromuscular problem, but is also determined by a cognitive–linguistic impairment. In support of this, pwP demonstrate difficulty, not just in monitoring and producing prosodic contrasts in their own speech, but they also demonstrate difficulty in perceiving these distinctions when spoken by others (discussed in the following section) [21–22,37–40].

▪ Language

Earlier views only considered language changes in PD in relation to dementia. It is now acknowledged that pwP experience problems in aspects of language processing that are independent of dementia, up to 50–60% in some estimates [6,13,28,41]. Whether these changes are specific to language processing or epiphenomena of more general cognitive–neuropsychological alterations or early precursors of dementia remains to be seen. The weight of argument currently rests with an interpretation of more general cognitive changes being responsible. Effects of bradyphrenia, deficits in verbal working memory, problems with set-shifting, poor attention and distractibility, dual-task processing factors, executive dysfunction, among others, are all cited as possible associates of language difficulties [42–53].

Regarding standard aphasia batteries and offline tasks (e.g., grammaticality judgement, lexical decision), pwP perform as well as matched controls (reinforcing the view that language changes, certainly in earlier stages, are secondary rather than primary, and that the employment of standard aphasia tests suggested to be prevalent in some surveys of practice is misplaced) [54]. Despite this, pwP may use shorter, less complex sentences and produce more grammatically wrong or incomplete utterances compared with matched controls [28,41,43]. Difficulties are reported with decreased semantic competitive inhibition, leading to an impaired ability to suppress unwanted material during lexical retrieval [45,46]. Lexical processing has been found to be worse for verbs than for nouns [44,49,50].

Similar to matched controls, pwP experience more difficulty processing longer and more complex stimuli. However, pwP perform more poorly on processing noncanonical sentence structures such as object-relative embedded clauses (e.g., the boys whom the pipers saw baked moist pumpkin pies) compared with subject-relative embedded clauses (e.g., the boys who saw the pipers baked moist pumpkin pies) or less complex/more canonical structures (the messy boys and the merry pipers baked moist pumpkin pies) [28,48].

Much meaning in language is signaled through prosodic variation rather than explicitly stated through grammar or word choice; for example ‘oh, you’ve finally arrived’ exclaimed with a note of relief versus annoyance. The likely culprit here for the difficulties pwP experience is the proposed prosody comprehension problem that has been linked to difficulty comprehending and expressing irony, sarcasm and humor [21,37–39,55]. The reported problems in resolving lexical and syntactic ambiguity (e.g., ‘they saw the boy with their binoculars’ or ‘they are hunting dogs’) and comprehending metaphor or figurative speech, (e.g., ‘her home was a prison’ or ‘he smelled a rat’) most probably also relate to this. These reflect the statements of carers who report that their partner fails to pick up on inferences from tone of voice, gets the wrong end of the stick (another figurative expression) in conversations and/or no longer has the same sense of humor [56].

▪ Nonverbal: gestural communication

Arm movements and facial expression contribute to understanding a speaker. Interlocutors also judge an individual’s mood and other qualities based on these signs [57]. PwP show reduced arm gestures accompanying speech and hypomimia is a well-attested feature [58]. These have been associated with reduced comprehension of pwP by others as well as with misperceptions of their affective state and general character [59].

▪ Pragmatics

Pragmatics is used here as an umbrella term to cover verbal and nonverbal aspects of handling conversations. Pragmatic difficulties are a source of communication breakdown for pwP on a number of fronts. Speakers employ a variety of nonverbal messages (e.g., facial expression, body posture and arm gesture) to signal their intention to enter a conversation and to hold their place once involved. As highlighted in the previous section, alterations to nonverbal gestures hamper pwP here. Subtle prosodic changes signal intention to carry on or to relinquish a turn, a signaling system that, as was noted above, suffers owing to the prosodic disturbance in PD. Having gained the floor, listeners expect one to commence a turn right away and interpret silence then or during a turn to signal an unwanted turn or its termination. The difficulties observed earlier for pwP in initiating and sustaining voice, and slowness of mentation in commencing and conveying a message, further disadvantages them here, as does the lack of prosodic cues signaling that their ‘turn is still open’.

Conversations entail rapid switches – for example, topic to topic, speaker to speaker, especially in multispeaker situations. PwP are impaired on certain aspects of switching tasks, especially if self-initiated [48,60–62]. Conversations demand focused attention, imperviousness to distractors in the external or internal (e.g., thoughts and associations) environment. Once more, pwP are taxed in these areas, as they are by the challenges conversations pose for constant problem solving at speed [28,63]. Added to this, conversations require constant dual-task/divided-attention performance (e.g., interpreting what has been said while formulating a reply, trying to gain entry into the conversation and focusing on the effort to maintain loudness and intelligibility).

If the notion of a receptive problem recognising gesture and prosody is upheld, pwP may experience barriers in conversations from missing or misinterpreting the gestural and prosodic cues of others concerning turn taking, monitoring for misunderstandings and managing repairs.

Indeed, indications suggest that the difficulty here is not purely a side effect of other changes associated with PD, but that there may be a specific problem with pragmatics [64]. PwP have difficulty interpreting speech acts. Others have hypothesized that this may be a manifestation in pwP of impaired theory of mind, and of interpreting the intentions and affective state of others [8,56,65–67]. Currently, the claim remains contentious.

Management

The UK NICE guidelines for PD state that speech–language therapy should be available throughout the time course of the condition and should address pitch, intonation and intelligibility [201].

Given that current interpretations of speaker performance suggest that the underlying source of difficulty in maintaining viable speech intelligibility does not arise from primary motor disturbance, but from impaired attention-to-effort/internal scaling and monitoring of output, then interventions that target articulators (e.g., tongue or lips) or specific sounds in isolation are unlikely to succeed. Indeed, no systematic reviews in the past have identified any therapies of this nature that benefit pwP and no studies supporting such regimes have appeared recently.

Conversely, attention-to-effort treatments and rate-control interventions that target (re)organization across the whole speech production system simultaneously have long been therapies of choice for speech clinicians in managing pwP, and recent theoretical models and reviews support their application [68,69].

Attention-to-effort therapies center on instructions such as ‘maximize your effort’, ‘say it as loud as you can’, ‘think loud’ and ‘internalize the sensation of what it feels like to produce speech that listeners judge as loud enough.’ This technique has, in recent decades, been formalized and marketed under the label LSVT, with accumulating evidence [24] that this brings about a significant increase in loudness and clarity of articulation [70]. Gains are in the region of 10–15 dB at the end of the 4-week intensive phase (16 face-to-face hours plus homework) with approximately 3–5 dB increases maintained at 12 and 24 months [71]. How far these gains translate into sustained improved intelligibility and whether success levels are uniform for all patients remains unclear [72]. Trials are underway at various centers to compare LSVT with other treatments and in less selective populations than has happened in the past. Exploratory work is also underway to examine LSVT efficacy after deep-brain stimulation (DBS) [73].

One problem regarding service delivery has been access to clinics/clinicians on an intensive basis. Recent findings suggest LSVT delivered remotely is not inferior to face-to-face sessions [74], and preliminary data established that an 8-week less-intensive version was also not inferior [75].

The success of attention-to-effort therapies is believed to derive from the fact that they target the central initiation and scaling impairment underlying voice–speech production; they are intensive and involve exhaustive repetition of target exemplars, they incorporate elements of training self-monitoring and readjustment of internal sense of effort and are not dependent on external feedback in the long run. They are delivered with a single instruction (e.g., ‘think loud’) and feedback is based on a single outcome (loudness) [24,68,76]. These motor learning elements appear superior over therapies in which more complex instructions and feedback challenge the attentional and dual-task difficulties of pwP, and where dependence on external feedback does not lead to internalization and generalization of new learning. An active ingredient may also be that the intensive work entailed in attention-to-effort therapies constitutes physical exercise, shown to benefit mood and cognitive functioning in pwP [77]. Various techniques for speech-rate control have also been gainfully used with pwP to improve intelligibility [78,79]. The precise mechanisms of effect here and for whom precisely different techniques are best suited are not fully elucidated [17,26,68,78–80]. Added clues for listeners regarding word and syllable boundaries doubtless represent one element; permitting more time for an underfunctioning system to realize goals and for listeners to process degraded sound signals are also likely factors. Imposition of a pacing cue may also be a factor in motor control reorganization, similar to outcomes seen in gait rehabilitation [81].

Recent investigations confirm earlier work that suggests that delayed- or frequency-altered feedback to improve rate and articulation control in pwP is not universally effective, though they are beneficial to some individuals [82,83].

One other approach to therapy explored recently concerns repetitive transcranial magnetic stimulation. The worth of this beyond placebo effect remains undecided, but work here is in its infancy in PD [84,85].

Little work has investigated the nature of communication changes in the late stage of PD and what may be the optimum speech–language intervention. Provision of high and low technology alternative and augmentive communication devices (e.g., voice synthesizers, signal ‘cleaning’, predictive keyboard systems, ABC charts, picture pointing and voice amplifiers) has been the principal practice to date [86,87], but there is clearly scope for much more preparatory work earlier on and more specific support work at the palliative stage [88].

The management of language and pragmatic changes has not been systematically studied. Analogous to similar changes found after stroke and head injury, most speech–language clinicians employ strategies of listener education (i.e., the nature of the problems, how to constructively support communication, how to negotiate repair of communication breakdowns and being a receptive and perceptive listener) and environmental manipulation (e.g., focusing of attention, removal of distractors) [54,68,89]. The field is rife for investigation.

▪ Pharmacological interventions & communication

In contrast to the demonstrated advantages for limb motor control from medical interventions, there is sparse evidence that they create similar gains for speech. Differences in methodological rigor and divergences in who has been examined, under what conditions and using which measures preclude firm generalizations. Recent reviews point to improvements in some underlying speech–motor parameters (e.g., tempo, gain settings and voice fundamental frequency), but no significant and maintained clinical or functional improvement in communication are reported [90–94]. Especially in the early stages, individual variability in pathological progression and dose sizes required to bring about limb motor improvements in contrast with cognitive changes may lead to relative flooding with L-dopa that may depress speech output and awareness of speech performance. Some cognitive processes are more closely tied to dopaminergic networks and so may be expected to improve (at certain disease stages) with dopaminergic therapies, while other processes see no change or decline [95–98]. Such findings support the possibility that reasons for the lack of efficacy for speech from pharmacological input are that speech–motor control is dependent on different networks to limb motor control and/or the cognitive components in speech–language control are not influenced by dopaminergic therapy.

Nevertheless, at an individual level (in contrast with group data) there are speakers who appear to derive significant communication benefit from medical therapies. A fruitful line of research therefore, would be to investigate in what way(s) these individuals differ from those who experience no benefit.

▪ DBS & communication

A similar situation to medical therapy pertains when attempting generalizations concerning effects of DBS on speech. There is no unequivocal evidence that changes extend beyond effects on isolated speech–voice parameters to encompass improvement of overall speech intelligibility, and some studies indicate deterioration in communication with DBS [99–106]. However, the issue remains far from settled. Small adjustments in the precise placement of electrodes and dose settings can result in differences to motor and cognitive outcomes [107–112] and individual patients may react differently to changes [113]. Tasks employed to measure outcomes also affect results [104,114]. It may be that pwP will need to make decisions regarding trade-offs between decreased dyskinesias, improved limb control and speech intelligibility.

Conclusion & future perspective

Communication changes affect the vast majority of pwP. Even though listeners may not hear these changes, pwP report them from the earliest stages and instrumental investigations illustrate that the changes are real.

Identifying all changes that may contribute to altered communication (i.e., language and pragmatics, not just the narrower attention to isolated voice and speech perspective that typified earlier research) demands comprehensive assessment at regular intervals across the time course of the condition. Given that psychosocial aspects of communication (e.g., loss of confidence, embarrassment, impact and so forth [6,7]) are significant concerns for speakers and carers alike, particularly in the early stages, assessments must cover activity, participation and impact aspects of functioning, not merely impairment dimensions. In parallel, interventions should also encompass these dimensions. As solutions to many difficulties that can arise can be better instigated before the changes become insuperable and before the pwP and their social circle adopt maladaptive strategies, timely, early referral to speech–language clinicians is recommended. Claims that certain types of speech intervention have neuroprotective properties, if confirmed, strengthen the notion of key early referral [24,69]. As the nature of issues in communication evolves over the course of PD, it also follows that disease assessment and management must reflect this changing picture.

Effective behavioral interventions exist that improve speech and voice, especially when delivered intensively and when focusing on effort and rate. An almost universal finding, not confined to the speech–language pathology literature, is that there is wide individual variability in response to different interventions. Further research must establish which pwP and which speech–language profiles respond best to which interventions and when. Variability may relate to well-known subtypes of idiopathic PD and the presence of atypical PD features [115,116]. Gender, cognition and other disease-specific and demographic variables are also likely to exercise an influence, alongside the fact that rate and profiles of change vary across individuals [5].

To date, behavioral treatments appear to offer better scope for speech than pharmacological or surgical interventions. The coming years promise to alter this. The technique of transcranial magnetic stimulation that has shown promising results in other neurological disorders and motor functions may offer new possibilities for improved speech–motor function in PD. This remains to be seen. More certain is that developments in electrode placement and dose control in DBS offer important avenues for further progress. Speech clinicians and speech scientists stand in an ideal position to offer sensitive measures within a multidisciplinary collaboration to determine which placements and doses influence which speech–language parameters, and to what effect as regards to intelligibility and understandability. The past decades have seen a refinement of pharmacotherapeutic interventions, but are still largely directed at symptom control. It is likely that we are at the threshold of developing drugs that not only control or retard symptom development, but can arrest symptom development or even reverse decline. Whether these new departures will also entail pharmacological and DBS therapies that can selectively target brain networks and processes that support voice, speech and language variables is an exciting challenge. In the meantime, behavioral interventions will remain an important aspect of PD management. Given the suggestions that speech–physical exercise may impart neuroprotective qualities, it is not beyond the bounds of possibility that the worth of such interventions may be felt well beyond the bounds of communication.