Speech Therapy and Neuroplasticity: How and why speech therapy increases fluency

Speech Therapy and Neuroplasticity: How and why speech therapy increases fluency This post follows [my previous post](https://www.reddit.com/r/Stutter/comments/pl522u/neuropathology_why_stutterers_brains_produce/). That chapter is much longer (a 45-minute read) but I'd recommend reading that first, since it explains the speech system better than this chapter, which is more about how speech therapy changes how the speech systems of stutterers. This section also not as polished as the previous, so you might have to be a little patient. :) The chapter after this goes into detail into how to make the most of speech therapy (or self-directed analogue). That will come later this week. If you want to read the actual studies, I'd recommend starting with [Kell 2009](https://academic.oup.com/brain/article/132/10/2747/330765?login=true). It may be difficult to make sense of the neuroscience, but it's one of the most accessible studies. # Speech Therapy and Neuroplasticity We know that the speech systems of stutterers have defects not seen in fluent speakers, and those flaws interrupt the smooth production of speech. By that same logic, if those differences were normalized, then the speech system of stutterers would function the same way as those of fluent speakers; presumably, stutterers would become fluent as a result. Is it possible to induce neurological changes - through a process called "neuroplasticity" - to correct the firing patterns that cause stuttering? Modern speech therapy aims to increase fluency by teaching new speech behaviors. First, we have to ask whether speech therapy effectively increases fluency at all. If speech therapy does work, then what effect does it have on the particular neural mechanisms that cause blocks? Could the conscious techniques taught in speech therapy change the subconscious activation patterns of the speech system? Separate studies conducted by Luc De Nil, Anne Giraud, Chris Kell, Chunming Lu, Katrin Neumann and Akira Toyomura scanned stutterers' brains before and after participants completed intensive speech programs. Each study employed different scanning technologies and slightly different speech interventions\*, but the general results were similar enough that they can be explored as a unit. \*(Lu and Toymura both used interventions other than traditional speech therapy.) These studies demonstrate that the speech systems of stutterers are not static. There are some core etiological deficits that seem to be irreconcilable, though the speech system will compensate for those deficits, for better and for worse. Speech therapy can reliably increase fluency, and it does so by normalizing many of the neurological differences between stutterers and fluent speakers. **Attempted Compensation** Participants in these studies were moderate-to-severe stutterers who had not been to speech therapy in at least three to five years, if ever; this gap in time allowed the researchers to observe a "raw," unguided response to the etiology of stuttering. The pre-therapy scans found the participants' speech systems were impaired not only by the original etiology of stuttering, but also by attempted compensations. The etiology caused repeated breakdowns in the speech system, so the brain tried to adapt around those weak points; however most of these adaptations only exacerbated the problem. A major causal factor of these maladaptations was the unreliability of auditory information. Recall that speech is processed primarily by the left auditory cortex, and the auditory signals must be processed by the motor cortex in order to be decoded as speech. However, in stutterers the left auditory cortex is liable to intermittently drop out, and the connection between the auditory and motor cortices is often impaired. This leads to gaps in incoming auditory information and delayed auditory processing, deficiencies that affect the speech system's ability to generate motor commands. While a combination of auditory and somatosensory feedback is needed to generate motor commands, auditory feedback is more critical/takes precedence. Auditory feedback controls the rhythmic flow of articulation, which is critical for fluent speech. When auditory feedback is impaired or unreliable, the speech system comes to stop expecting it. This causes the speech system to over-rely on somatosensory feedback, to the point that the auditory-somatosensory balance tips to the latter. While most speech processing occurs in the left hemisphere of the brain, the brain can react to deficient processes in the left hemisphere by recruiting mirror components from the right. These homologues are roughly similar in functionality, so they can contribute processing power, but it's a less-than-ideal solution. These homologues are not specialized for language like the original components, nor are they as physically close to the other components of speech production; their signals must travel farther and pass through the relatively narrow connection between the two hemispheres. Studies suggest that this interhemispheric connection - the corpus callosum - is impaired in stutterers, further exacerbating the issues caused by relying on homologues. Chris Kell observed that stutterers had motor over-activations in the right hemisphere which correlated with dysfluency. These overactivations could simply be the result of poor speech planning - incomplete/unreliable speech plans required extra motorific effort to put into action - or they may have been attempted corrections to motor control errors. Even if they do not directly cause dysfluency, it is more noise in a system that is already impaired. Why are these attempted compensations bad? Could these workarounds be more effective than trying to correct the problems with auditory feedback? As Katrin Neumann referenced in one of her papers, in the case of damage caused by stroke, recovery that approximates "normal-like" brain activations indicates good functional recovery whereas more activation outside the default networks signal poor outcome. It's not all bad, though. Some unguided compensation attempts correlated with increased fluency. Brodmann's Area 47/12 of the right hemisphere contributes to the integration of auditory feedback into motor control. Stutterers have deficiencies in that functionality, so it is unsurprising that stutterers show increased activation of this area, and that activation correlates with improved fluency. Stutterers also exhibit increased activation in the right orbitofrontal cortex during speech. Andrew Etchell proposed that this activation counteracts some of the internal timing deficits seen in stutterers. Anne Giraud recorded activity in the right frontal operculum, which she ascribed to beneficial self-monitoring and language repair. Stutterers also show activation patterns in the cerebellum that distinguish them from fluent speakers. The cerebellum sits in the lower back-half of the skull and supplements timing, motor control, and self-monitoring. Multiple studies have found differences in how stutterers and fluent speakers activate the cerebellum during speech; while the results are not unified enough to definitively state how this activation impacts speech production in stutterers, it appears to have a beneficial effect on timing, motor control and self-monitoring. These findings, taken together, demonstrate that the brain's unguided response to the etiology of stuttering can lead to even worse dysfluency. That brings us back to speech therapy, which is a structured behavioral response to stuttering. What effect does speech therapy have on fluency and neural activation patterns? **Speech Therapy** In all of the studies introduced earlier, nearly all participants saw improvements in tangible speech metrics. Chris Kell's participants had their dysfluency decreased from 7.4% stuttered syllables to 0.6%. Similarly, Katrin Neumann reported that, after therapy, her subjects had better control of their vocal cords and more dynamism in their speech. Clearly, speech therapy was effective at improving fluency, but how did it affect the neurological processes of speech production? Therapy abolished the maladaptations and brought the participants' speech systems closer to those of fluent speakers. Kell observed that the connection between the auditory and motor cortices had been improved, leading to better processing of auditory input. Additionally, auditory feedback had been re-integrated into speech production and once again took precedence over somatosensory feedback. He also observed that the motor activations in the right hemisphere had been normalized, and most of the processing once again occurred in the left hemisphere. Improvements were also seen in timing functionality. Prior to therapy, Akira Toyomura observed below-average activation in the basal ganglia and above-average activation in an area of the cerebellum distinct from the speech network. (This cerebellar activation was likely compensating for deficient internal timing.) His intervention normalized values for both, raising activation in the basal ganglia and lowering it in the cerebellum. Though Toyomura did not directly measure beta oscillations in the basal ganglia, we can infer that these changes indicate improvement in the internal timing circuit. Anne Giraud observed that, before therapy, her subjects had abnormal activation patterns in the caudate, a part of the basal ganglia. Interestingly, some subjects had abnormally high caudal activation while others were abnormally low. Therapy, however, brought both groups to activation levels seen in fluent speakers. While most studies had subjects read neutral statements during scanning, prosody\* was a key factor in Katrin Neumann's study. Her participants read nonsense sentences\^ under three conditions: no affect (neutral prosody), as if they were conveying happiness (emotional prosody), or as if they were asking a question (linguistic prosody). Before therapy, stutterers matched LIFG activation in fluent speakers during neutral prosody - which requires relatively little LIFG activation - but under-activated the LIFG during emotional and linguistic prosody. Therapy normalized activation levels for emotional prosody, but not linguistic. \*(The rhythm and intonation of speech, used to convey the emotions and/or intent of the speaker.) \^(i.e. "Blue dwarves like to run through the grass.") Speech therapy techniques reduce blocks, but they also result in speech that often is monotonous, expressionless, unpleasant, and unnatural. Studies led by Marie Christine Franken, Roger Ingham, Joseph Kalinowski, and Andrew Stuart measured stutterers' fluency and speech naturalness after participants completed intensive speech therapy programs. In every study, even though therapy increased fluency, it also decreased naturalness. Even for severe stutterers, highly dysfluent speech (8-17% stuttered syllables) was rated more natural than near-perfect fluency after therapy. These losses in naturalness were somewhat ameliorated a few months later at a follow-up measurement. However, those gains in naturalness were accompanied by a small drop in fluency. That begs the question, will improved fluency earned during intensive therapy survive in the long run? Using therapy techniques for multiple hours a day creates neuroplastic change, but will those changes regress in the months and years that follow? **Follow-up and Maintenance** A 2009 study led by Lisa Iverach found that while therapy improved fluency, at a follow-up measurement six months later two-thirds of the participants had regressed to 5% stuttered syllables. However, these subjects were not enrolled in a rigorous maintenance program. How would such a program affect the persistence of fluency gains? Two such studies, one each by Luc De Nil and Katrin Neumann, enrolled their subjects in maintenance programs lasting one-to-two years after the conclusion of intensive therapy. In De Nil's case, sessions were geared around incorporating fluency skills into daily life as well as cognitive and attitudinal interventions. These sessions were held on a descending schedule; they took place once a week, then once every other week, and eventually once a month. These same subjects returned to the lab anywhere from one to three years later to redo the initial battery of speech evaluation and brain scans. The post-therapy peak of 0.6% stuttered syllables regressed to 2%, which was still much improved from the 6% prior to therapy. The neurological changes from therapy were maintained, and had even matured. The maladaptations resolved by therapy did not return and the study cohort still matched fluent speakers in auditory-motor connection and auditory-somatosensory balance. Neumann found that her participants still matched fluent speakers' LIFG activation patterns in neutral and emotional prosody, and had even developed normalized activation during linguistic prosody. **Why It Works** The therapeutic programs used in these studies were not developed in response to the findings of modern neuroscience. Fluency-Shaping Therapy and Kessel Therapy were developed in 1974 and 1980; neuroplasticity is a relatively new concept, and nearly every study mentioned in this book was published after the year 2000.\* It's incredible that these speech therapists developed an effective way to create neuroplastic change before they knew the specific neurological deficits of stuttering, or that it was even possible to change them. Regardless, it's clear that intensive speech therapy reliably stimulates neuroplastic change that normalizes many of the deficits in stutterers' speech systems. \*(The earliest neurological study of speech therapy in this chapter was published in 2001.) One clue as to how this occurs comes from the work of Oren Civier and GODIVA. As explored in the previous chapter, Civier could make the GODIVA model "stutter" when he altered the program's parameters to match deficits in stutterers' speech systems; namely, weak articulatory motor control and delayed processing of auditory feedback. Impaired articulatory control caused the model to miss speech targets, and the delayed processing of auditory feedback prevented the auditory error maps from recognizing and correcting for those errors. Therefore, as the model transitioned from phoneme-to-phoneme and syllable-to-syllable, errors carried over and compounded until the model was so far off-target that it triggered hard blocks and repetitions. Civier ran the model again with the same parameters, but this time he had it mimic a speech therapy technique called "elongated syllables." When using this technique, the speaker slows their rate of speech by holding each syllable twice as long as they normally would. Civier demonstrated that the slower rate of speech allowed more time for delayed auditory processing to occur. With auditory feedback re-incorporated into the speech system, the model was able to recognize minor errors; with more time to produce each syllable, the error maps were better able to correct these errors. As a result, the model was significantly more capable of reaching the full and accurate expression of each phoneme. Reaching the target for each phoneme led to more accurate transitions, which led to increased fluency. Elongated syllables increases connection to auditory feedback while it's being used, but it may also increase that connection during regular speech. Kell noted that, prior to therapy, stutterers had a healthy ability to detect "slow modulations in acoustic speech parameters" but weak ability to detect rapid changes in the same. Therapy resolved the weak connection to rapid modulations. Therefore, one could postulate that tuning into the slow modulations of elongated syllables brought auditory feedback back into speech production; this may have then re-ignited the fast-modulation system, which is highly applicable to monitoring auditory feedback of speech. Kell postulated that conscious use of speech therapy techniques - rather than rote speech patterns - would require more speech planning, which counteracts deficient speech planning in stutterers. He also speculated that slowing down speech to produce elongated syllables requires the speaker to be more conscious of speech timing, which would counter potential negative effects of the internal timing deficit. **Conclusion** When examined in total, these studies paint a pretty clear picture: intensive speech therapy followed by maintenance programs increases fluency by normalizing both causal deficits and second-order adaptations in the neural production of speech. There are other interesting points to note. These studies show that intensive speech therapy can tame even the most severe stutter. However dysfluent participants were at the beginning of therapy, nearly all finished the program speaking with 0-2% stuttered syllables. This was true even for the stutterers in De Nil's 2003 study who entered the program speaking with 11-23% stuttered syllables. It's important to note that all of these studies used intensive programs; I didn't find any studies of programs with sessions that met once a week, or less. It appears that intensive therapy programs stimulate more neuroplastic change than less-intensive programs of equal hours, so it's reasonable to suppose that weekly-based programs would be less effective at creating neuroplastic change. There is plenty of reason for optimism when it comes to speech therapy. This much was developed without knowing the neuroscience of the underlying cognitive causal factors nor how therapy corrected them. It will be interesting to see if these neuroscientific advances can be incorporated into making speech therapy even more effective. As time goes on and brain-scanning technology continues to improve, we should learn even more about the effects of therapy.