On Target Practice and Speech Therapy

Research Digest (http://bps-research-digest.blogspot.ca/2012/08/targets-look-bigger-after-shot-that.html ) posted a rather curious article about the relationship between perceived target size and shooting accuracy. In the context of archery, larger targets are easier to hit; at the same time, when the target is successfully hit the archer retrospectively perceives the target to be bigger than it was. Apparently this works for golf too in the context of putting and perceived cup size. The authors of the golf research suggest a reciprocal relationship between perception and performance. I don’t recall ever noticing the retrospective relationship in my own life but I do know that prospective perception of the target is a critical factor and that my golf game has always been hampered by poor representation of target location. It is very annoying how often my husband tells me “that ball went right where you aimed it” just as it is disappearing into the forest far from what I thought was my target.

Last week I spent more time in the lab watching some of the 1000 hours of video we recorded as part of the ECRIP trial (Essai Clinique Randomisé sur les Interventions Phonologiques; a randomized control trial of phonological interventions for francophone children). It was a great pleasure to watch one of our most talented students teaching two delightful children to articulate the /ʃ/ phoneme using one of the more popular games in the Output Oriented treatment arm. The game is called ‘course des animaux” (animal race): it provides opportunities to practice animal names that include the /ʃ/ phoneme in different word positions such as “chat” (cat) → [ʃa], “cheval” (horse) → [ʃəval], “cochon” (pig) → [kɔʃɔn], and “vache” (cow) → [vaʃ]. One child in particular was having a lot of trouble with “vache”, producing [vas] or [ʃvas] on each spontaneous trial although achieving it with pictogram support to help him sequence the head and coda of the word appropriately. The student clinician was providing excellent feedback and prompting the child to use the pictographic cues at appropriate intervals. What was clear from the child’s face however is that he had no idea of the target; he had to look to the student clinician on each trial to verify if he had achieved it. Whether he said [vas] or [vaʃ] he would produce the trial with the same frown of concentration and then look to the clinician for feedback; if she indicated that he had produced the word with “un beau rond” (a good circle), only then would he break into a smile. Even more interesting, another student arrived in the lab while we were watching the video and informed us that this child was related to her and that he was still misarticulating “vache” in spontaneous conversation, now two years after the videorecorded therapy session (it takes us a long time to analyze our data!).

We can contrast this exchange with another one that is reproduced (as a hypothetical English version of the original French session) in Chapter 9 of Developmental Phonological Disorders: Foundations of Clinical Practice. The exchange involves a child who was assigned to the Input Oriented treatment arm of the ECRIP trial. All the Input Oriented treatment procedures were directed at strengthening the child’s acoustic-phonetic or underlying phonological representations of the target form. The sessions are markedly different from those conducted in the Output Oriented intervention in which the child is practicing speech all the time; in the Input Oriented sessions the focus is on the therapist providing input while the child listens. The procedures that were used in the Input Oriented intervention were as follows: In the first session the student therapist would identify the target for the child. Then the therapist would proceed to auditory bombardment in a meaningful context such as a story. For example, when /ʃ/ was the target, Michel, le mouton qui n’avait pas de chance (Sylvain Victor)was a favourite for our student therapists to read since it is a delightful story that provides at least one /ʃ/ exemplar approximately every three seconds of reading. Auditory bombardment would be followed by activities in which the child actively judged words to be correctly or incorrectly pronounced. This is the specific exchange that I want to contrast with the “course des animaux” activity. Rather than asking the child to name animals as they participated in a race to the farm, the therapist essentially asks the child’s permission to put the animals in the barn – he has control of the animals but they cannot go in unless the therapist pronounces the words correctly. While he holds the horse, she says to the child “J’aimerais faire entrer le [sival] (I would like to make enter/let in the horse (pronounced incorrectly)”. The child tells her he won’t let in the horse because she must say it “like it should be said”. So she tries again, “J’aimerais faire entrer le [səval]”. Now the child gets impatient. He explains that he can’t let in the horse unless she says [ʃəval]. “Oh”, exclaims the therapist, “J’aimerais faire entrer le [ʃəval]”. Without wasting any time the child gallops the horse into the barn. At this moment the child has shown us that he has achieved perceptual and productive knowledge of the target form. Now we know that he is ready to progress to expressive minimal pairs therapy activities and the focus will switch to speech production activities. Up until this time there has been no explicit expectation that the child produce speech unless he wished to. If the child attempted a target word the therapist would respond with recasts, imitations and expansions but no explicit information about the child’s production accuracy. We haven’t finished our data analysis yet but we know that the Input Oriented intervention was at least as effective as the Output Oriented intervention on our single word picture naming test even though the children in this treatment arm received very little direct speech practice during their therapy sessions. Our preliminary results mirror those reported by Hesketh and colleagues in which they found that standard articulation therapy and a phonological awareness intervention had similar outcomes in the short-term but the phonological awareness program appeared to provide a better basis for spontaneous generalization by the children over the long term (http://onlinelibrary.wiley.com/doi/10.1080/136828200410618/abstract).

We think that Input Oriented therapy has the advantage, for all children, of providing them with a clear sense of the target so that they can determine for themselves whether they are achieving it or not. Children who receive standard articulation therapy tend to be very reliant on their therapist for information about their own speech accuracy which deprives them of hours of speech practice every day. Children who know what the target is can practice independently. This approach to therapy is also especially appropriate for children who cannot or will not engage in structured speech therapy activities due to their age or cognitive level or temperament. Finally we found that it was an excellent choice for children who had “bottomed out” with standard therapy and were in no mood to continue cooperating with speech practice after a long period of failure to succeed with a given target.

Francoise and I will teach a workshop at ASHA this year on how to implement Input Oriented therapy.

Topic Area: Speech Sound Disorders in Children (SLP)
Session Number: 1420

Title: Input-Oriented Approaches to Intervention for Children With Developmental Phonological Disorders

Session Format: Oral Session (Seminar 2-Hours)
Day/Time: Saturday, Nov 17 — 08:00 AM – 10:00 AM

Authors: Francoise Brosseau-Lapre, McGill U; Susan Rvachew, McGill U

On Birds and Speech Therapy

This past winter Kevin Munhall and colleagues at Queens University made a big splash when they concluded that toddlers do not monitor their own speech; for some reason this made the headlines in quite a few news venues starting with Science Daily (http://bit.ly/O45LDH). In their study they used one of the most fascinating procedures for studying feedback control of speech – a procedure that was pioneered by John Houde (http://bit.ly/Opqtnn) who, as an aside, recently published a superb article called “Speech Production as a State of Feedback Control” (http://1.usa.gov/Nf3k2v). As another aside I believe that the first published speech adaptation study with kids was actually done in my lab by Doug Shiller (http://bit.ly/PSlmXk) (http://bit.ly/xCdRsQ). In Munhall’s study, toddlers, children, and adults were asked to say the word ‘bed’ repeatedly while an apparatus shifted the first two formants of their speech closer together so that they would hear themselves saying ‘bad’. As expected, the children and adults compensated by changing their articulation of the word to move their formants even further apart. The toddlers did not compensate which is what prompted the authors to conclude that the toddlers were not monitoring their own speech. I think there are quite a few dodgy things about this study but I’ll get to them later because the most interesting part of the paper is the discussion in which the authors suggest that one explanation for the toddler’s failure to compensate can be found in the literature in vocal learning on birds. I was reminded of this study as I observed the birds at the feeder outside my window at the cabin where I am working. It was from the Munhall paper that I learned that there is fascinating bird research going on at McGill (strange I hadn’t heard of it before).

Vocal learning in song birds and humans share many similar processes but the processes are easier to pull apart and more amenable to laboratory investigation in the songbird (http://bit.ly/NjNAeJ) . Parts of the vocal learning process that seem all jumbled up in the human infant are more clearly separable in the bird because they occur in more specific times and places. First the young male bird, in a listening stage, learns the target song from his father. Later during the subsong phase the bird practices the individual elements of the song in a manner similar to the human babbling stage. This is followed by practice of the sequenced elements in a stage called plastic song. Finally crystalized song emerges and the bird uses this to attract females. Sakata and Brainard (http://biology.mcgill.ca/faculty/sakata/articles/Sakata_Brainard2009.pdf) from McGill explain that even adult finches practice their song when they are alone however (undirected song) and that this practice song is somewhat less structured than the song produced during actual courtship performance (female directed song). Now, the cool part is that Sakata and Brainard tested the influence of auditory feedback on the finches’ vocal motor control using a procedure rather similar to the one Houde designed for humans. And it turns out that manipulations of feedback of the birds’ songs influenced their singing during practice (undirected song condition) but not during performance (female directed song condition).

This reminds me of the way that toddlers will wake the whole house up happily yapping away to themselves early in the morning but later in the day when a cookie is desired, resort to whining and pointing.  For toddlers and for birds, vocalizing for practice and communicating to meet a functional goal are two very different contexts.  Feedback control of speech may be strongest under practice conditions. When the toddler wants a cookie, he or she may ask for one if a well-practiced motor plan for an approximation to the word is available but otherwise nonverbal communication will have to do. It is generally accepted that, once learned, sensory feedback is not essential to speaking. Once the toddler learns to say ‘cookie’ the motor plan for the word can be implemented with feedforward control. Sensory feedback is absolutely essential to the process of learning to talk however. If access to this feedback is attenuated in functional contexts, what are the implications for speech therapy?

Many approaches to speech therapy involve working with the child in functional, communicative contexts. Francoise and I describe these in Chapters 9 and 11 of our book. These approaches have the purpose of strengthening the children’s acoustic-phonetic representations for speech targets or helping the child experience the functional consequences of using the appropriate phonological forms in communicative contexts. When the child’s fundamental underlying problem is with processing or organizing phonological input these approaches are known to be effective and lead to rapid carry-over of learning to non-clinic settings. Throughout our book and in previous blogs I have stressed functional contexts as a motivator for speech practice and I have talked about authentic practice contexts as being essential for carryover. Is it possible that some aspects of speech motor learning are best learned in non-communicative contexts however?

During my 20 years of clinical practice, prior to switching to full-time academic work, I specialized in the treatment of childhood apraxia of speech. With these children I used a different approach designed to foster the development of speech motor control. Described as a sensory-motor approach in Chapter 10 of the book, this approach focuses on intense practice of nonsense syllable drills. Theoretically these children have a problem that is specifically with the development of motor plans for speech and thus access to sensory feedback during speech practice will be very important for the success of any therapeutic intervention. Possibly working with non-meaningful speech material is one way to help the child use critical sensory feedback for speech motor learning during speech therapy sessions. Neither birds nor humans practice in social isolation however and social cues are known to shape the quality of the bird’s song. Currently there is very little research to help us choose the right mix of drill versus play and nonsense versus meaningful practice material. There are two research programs we can keep an eye on though. Patricia McCabe and colleagues have registered a randomized control trial to compare their Rapid Syllable Transition Treatment (ReST) with the Nuffield Approach (http://bit.ly/Pbsxy7). One of the features of ReST is practice with multisyllable pseudowords in quite structured contexts employing principles of motor learning. I am very excited about the trial and look forward to seeing the outcome. I also plan to keep watching the birdsong literature because I think that there is a lot more to learn about the social, behavioral and neurophysiological conditions that promote vocal learning and I have a feeling that more useful discoveries will be coming our way from the bird laboratories in the near future.

Although I enjoyed Sakata and Brainard’s paper and think that it may provide some support for sensory-motor approaches to the treatment of CAS, I can’t see how their findings explain the toddlers’ failure to compensate in Munhall’s study however. The children and the toddlers produced the word ‘bed’ repeatedly in the context of a sort of video game in which saying the word bed caused an animated robot to move across the video monitor. I am trying to imagine why feedback control would become more rather than less active in that context as children become older and more practiced at speech. To be honest I am not convinced that the results have anything to do with the toddlers’ access to or use of auditory feedback. The paper lacks any perceptual evidence that the acoustic manipulation of the toddlers’ speech resulted in changes that were equally perceptible (compared to the manipulation of the adult’s speech). There are plenty of reasons to think that these manipulations would not be perceptible to the toddlers involving both the acoustic characteristics of the toddlers’ speech and known biases in toddler perceptual abilities. Secondly and more importantly the standard deviations of the toddlers’ formant frequency values during baseline was 100Hz and the expected shift in formant frequencies was 100 Hz! How did they expect to see a 100 Hz compensatory shift against a background of 100 Hz noise? I wouldn’t conclude that toddlers don’t monitor their own speech just yet. None-the-less, I enjoyed finding out about the birdsong research and the idea that access to feedback might be context specific is something to keep in mind.

On Golf and Speech Therapy

Last weekend I didn’t write a blog post because the weather was spectacular and I was having too much fun enjoying the new deck at my cabin and playing golf. This weekend it is raining so I have plenty of time to reflect on the decline in my golf skills since I gave it up three years ago to devote my weekends to writing a book. My daughter says I should go get a lesson but I am too embarrassed to do that because I am in such poor physical shape. I am not convinced it would help in any case because my husband and I used to go get tune up lessons in the spring and I was never convinced that these were a good investment. The instructors would give me a bucket of balls and leave me by myself while concentrating on my husband. This used to annoy me no end – I thought it was some sort of sex discrimination thing until I complained one time and the instructor says, “no, no – you really don’t need my help, you have a perfect swing, just keep practicing as you are”. Imagine my surprise! If I had a perfect swing, why was my score so awful (even before I gave up golf to write a book my scores were pretty awful but I had a terrific 200 yard drive so my score didn’t bother me so much). Anyway, when I was writing chapter three, I had to study up on theories of speech motor control and I figured out why my perfect swing wasn’t much good to me.

Practice Conditions

The problem is related to the vast difference in the practice conditions for golf relative to the actual playing conditions. Golf lessons and most practice sessions occur at a driving range or a golf dome as illustrated here: the terrain is perfectly flat and the practice mats are positioned to ensure that your body is aligned square to the target line. During practice it is common to hit many balls with the same club, concentrating on executing the same motor plan with a high degree of precision.

Play Conditions

Play Conditions

Our playing conditions are vastly different since our cabin is located 5 minutes from a course where, for $1500 a year the whole family can play as often as we have time for, with carts and no tee times – just show up and play nine holes when we feel like it, it’s wonderful. There is a hitch though and that is that it’s pretty much the worse golf course in the world except for the scenery. There isn’t a flat spot on it and that includes the tee boxes. Every time you hit your ball you are likely to end up with a bad lie like this one (ball below feet, basically hitting off gravel). All the precision in the world with my perfect swing is not going to help me hit this ball. What I need to do is process the initial conditions accurately and select a motor plan that is going to get the club to the ball given those conditions. Looking at the picture I can tell that my ball is too far back in my stance but at the time I was quite unaware that I had positioned myself incorrectly relative to the ball – often my problem is one of poor information processing that leads me to essentially select the wrong initial conditions for the purpose of predicting which motor plan will have the desired effect.  Poor execution of my swing is not the problem. Unfortunately when I achieve the inevitable bad result I start to adjust my swing which just makes everything worse. Instead, I need to focus on processing aspects of the context so that I can adapt my set-up to the initial conditions: First, what is the gradient of the slope between my feet and the ball? How much of an incline is there in the lie of the ball? Where is the target relative to the ball? And then, where are my feet relative to the ball? How wide is my stance? How bent are my knees? Where is my centre of gravity? Are my shoulders aligned with the slope of the hill?

So what has golf got to do with speech therapy? Given that speech is also a motor learning problem, the same principles of motor learning that apply to golf apply to speech learning. I have spent a bit of time this summer watching therapy videos as part of the treatment fidelity process in the randomized control trial that Françoise and I recently completed. I see student clinicians and sometimes the experienced SLPs conducting therapy sessions like practice at the golf dome – aiming for precision rather than dynamic stability. I think the students should know better because I taught them the principles of motor learning using Maas et al.’s (2008) excellent Tutorial http://bit.ly/Ta9STv in which the authors stress the distinction between learning and performance. Performance during practice may or may not transfer to untrained movements in nonpractice contexts. Maas et al. discuss a number of different strategies to enhance transfer of training to similar but unpracticed movements. Although the research findings are complex and often difficult to interpret, it appears that overall it is best to practice under conditions that afford a wide range of experience with varying initial conditions and movement outcomes. At the golf dome the best one can do is switch clubs and targets often. In speech therapy, practicing the target phoneme in many different words so as to vary phonetic context is often a good strategy. I think that novice and experienced SLPs know that variable practice conditions are important but it is not always easy to implement this principle for two reasons. The first is that performance levels are higher under constant than variable practice conditions and it is reinforcing to both clinician and patient to achieve high levels of accuracy during therapy sessions (the distinction between practice performance and actual learning is hard to keep in mind). The second is more fundamental: the goal of the therapy exercise is not itself clear to SLP or patient. In some ways, Maas et al.’s Tutorial contributes to this confusion of aims by focusing on motor programming and motor programs. Therapy sessions are conducted as if the goal is to perfect the specification and execution of a particular motor program. I prefer Wolpert’s approach to motor learning http://bit.ly/OE8VT5 (take a look at this if only for the Calvin and Hobbe’s cartoons). Wolpert and colleagues (2001) describe motor control “as the process of transforming sensory inputs into consequent motor outputs. The problem of motor learning is one of mastering and adapting such sensorimotor transformations” (p.488). We can think of speech therapy as the process of helping the patient to process the sensory input so as to transform them into the desired motor outputs. An approach to motor learning that takes into account information theory and information processing is the “challenge point framework”, described by Guadagnoli, M. A. and T. D. Lee (2004) http://www.tandfonline.com/doi/abs/10.3200/JMBR.36.2.212-224#preview. Françoise and I are going to teach a seminar about how to apply this framework to speech therapy at ASHA 2012 in Atlanta this fall:

Topic Area: Speech Sound Disorders in Children (SLP)
Session Number: 1530

Title: Application of the Challenge Point Concept to Developmental Phonological Disorders

Session Format: Oral Session (Seminar 2-Hours)
Day/Time: Saturday, Nov 17 — 03:00 PM – 05:00 PM

Authors: Susan Rvachew, McGill U; Francoise Brosseau-Lapre, McGill U