Never Segment Words in CAS? Then What Do You Do When the Child Can't Say the Word?
Updated: Sep 23
One of the most common pieces of advice clinicians hear in Childhood Apraxia of Speech (CAS) is:
"Never segment words."
The concern is understandable.
We don't want children producing robotic, choppy speech.
We want smooth, connected movement.
We want coarticulation.
We want natural speech.
I want all of those things too.
But here's the question I often get asked when providing CAS clinical supervision:
What do you do when the child can't say the word?
You present the whole word.
They can't do it.
You slow it down.
They still can't do it.
You stretch it out.
Still can't do it.
Now what?
At this point, many therapists feel stuck.
If you can't make the task easier, you are left with only two options:
1. Keep asking the child to attempt a movement they cannot yet produce
2. Or abandon the target altogether.
Neither option is particularly helpful.
And that's where I think we need a more nuanced conversation about segmentation.
After more than 20 years working with children with CAS, this is probably the single biggest lightbulb moment I see when supervising speech pathologists.
When clinicians learn how to break words down in the right way, their success rate often changes dramatically because they finally have a way to make a previously impossible task achievable.
The Problem Isn't Segmentation
I don't think the issue is whether we segment.
I think the issue is where we segment.
When many people hear the word segmentation, they imagine something like:
b - e - d
p - u - p - p - y
In other words, breaking the word into individual speech sounds.
For a child with CAS, that can actually increase the motor planning demands.
Instead of organising one coordinated movement, they now have to organise multiple isolated movements and somehow stitch them back together.
That isn't making the task easier. Often it's making it harder.
CAS Is a Movement Disorder
CAS is not primarily a disorder of sounds.
It is a disorder of planning and programming speech movements.
That distinction matters.
When a child cannot say a word, my first question isn't:
"What sound is wrong?"
My first question is:
"What movement is breaking down?"
Once you start viewing speech this way, the syllable suddenly becomes much more important.
Why I Protect the CV Unit
One of my guiding principles is to preserve the consonant-vowel (CV) movement unit wherever possible.
Instead of:
b - e - d
I might use:
be - d
Instead of:
p - u - p - p - y
I might use:
pu - ppy
One reason this matters is that simply slowing a word down doesn't necessarily make it easier.
If a child can't say puppy, stretching it out as:
ppppuuuuppppyyy
doesn't actually change the motor plan. It just makes the child perform the same difficult motor plan more slowly.
But if we change it to:
pu - ppy
we've actually changed the task.
We've reduced the motor planning demands while preserving the syllable-level movement patterns that speech is naturally organised around.
That means when we eventually remove the support, pu - ppy can naturally become puppy.
In contrast, if we break the word into isolated sounds:
p - u - p - p - y
we've dismantled the very movement pattern we're trying to teach. The child is no longer practising a simplified version of puppy. They're practising a sequence of isolated sounds that must later be reorganised into a completely different motor plan.
But Why Not Just Slow Down or Choose an Easier Word?
I absolutely do both.
Slowing the rate, elongating sounds, simultaneous production, forwards and backwards chaining, and choosing simpler targets are all valuable strategies. We can also temporarily remove phonation and focus on the articulatory movements alone, before introducing whispered productions and gradually returning to normal voicing.
If a child can successfully produce the word using these supports, there's no reason for me to segment it.
But what happens when these strategies still aren't enough?
We could continue working on simpler syllable shapes, and sometimes that's exactly what the child needs. However, for children with severe or complex CAS, this can leave us with a very limited selection of functional words.
That's where I find syllable scaffolding particularly useful. By temporarily simplifying a word while preserving accessible consonant-vowel movement sequences, we may be able to help a child produce a meaningful approximation of a word that was previously out of reach.
This isn't about replacing work on coarticulatory transitions. It's about finding an achievable starting point from which we can build towards them.
Interestingly, Speech Development Seems to Agree
What I find fascinating is that this isn't just something I've observed clinically.
Several areas of research point in a similar direction.
Developmental researchers have long noted that early speech is dominated by simple consonant-vowel syllables such as:
mama
baba
dada
MacNeilage and Davis proposed that these early CV productions reflect the natural organisation of the speech motor system, with rhythmic opening and closing jaw movements creating consonant-vowel combinations as foundational speech units.
Infant speech perception research also suggests that syllables are highly salient units for young children. Long before children become aware of individual speech sounds, they appear sensitive to syllables and larger speech chunks.
In other words, both speech production and speech perception seem to give syllables a privileged role.
What Does the Research Tell Us About Syllable Structure in CAS?
Interestingly, research involving children with CAS points in a similar direction.
Jacks et al. (2006) followed three children with CAS over three years and found that consonant accuracy was associated with syllable-shape accuracy and the frequency of multisyllabic productions. Their findings suggested that some consonant errors may be related to difficulties constructing syllable structures.
The researchers also identified syllable construction as a potential therapeutic goal.
And this is where things get particularly interesting clinically.
Of course, children need access to the individual consonants and vowels to construct syllables. But having those sounds available doesn't necessarily mean they can successfully produce them in every syllable shape.
When the syllable structure exceeds a child's current motor planning abilities, even a consonant they can produce accurately in a simpler word may be deleted, distorted or produced completely off target.
That's why I don't just ask whether a child can produce a particular consonant. I also want to know which syllable structures they can successfully construct, and at what point increasing the complexity causes their speech to break down.
Speech Motor Control Research Also Raises an Interesting Possibility
One of the most fascinating papers I've read recently wasn't actually about CAS.
It was trying to answer a completely different question:
How does the brain manage the enormous complexity of speech production?
To produce even a simple syllable such as ma, multiple speech movements have to be coordinated almost simultaneously. The lips, jaw, tongue, velum, respiratory system and larynx all need to work together in a precisely timed way.
Rather than treating speech as a string of separate sounds, Yi Xu proposed that the speech system may reduce complexity by synchronising consonant and vowel movements into a coordinated syllable-level unit. In other words, the syllable itself may be one of the mechanisms that makes speech possible.
What surprised me most was where the theory ultimately landed. After attempting to solve a speech motor control problem, it arrived at the conclusion that:
The CV syllable may be one of the fundamental organising units of speech.
Rather than viewing speech as a series of independent sounds that are later assembled, the paper presents evidence that consonant and vowel gestures begin together and function as a coordinated movement unit from the onset of the syllable.
As I was reading it, I couldn't help thinking:
Of course.
That's exactly what many of us see clinically in CAS.
The issue often isn't that the child doesn't know the consonant. Or the vowel. It's that they can't reliably coordinate the movement between them.
And perhaps that's why preserving syllable-level movement units can be so powerful.
When I use:
pu - ppy
I'm still working with the movement units that speech appears to be organised around.
When the support is removed, that movement naturally flows back into puppy because the underlying organisation of the word has been preserved.
When I use:
p - u - p - p - y
I've dismantled that organisation completely.
What I found particularly interesting is that computational simulations discussed in the paper suggest that speech learning becomes more efficient when these synchronised syllable-level structures are preserved. The authors describe successful artificial vocal learning models built around this principle of reducing complexity through synchronised speech movements.
I'm certainly not suggesting we should base therapy on artificial intelligence.
But I do find it fascinating when speech motor control research, infant development, speech production theory and computational models all seem to point in a similar direction:
The syllable may not simply be a convenient way of describing speech. It may be one of the fundamental ways speech is organised.
Syllables Don't Disappear Once We Learn to Talk
What I find particularly interesting is that the importance of syllables doesn't seem to disappear once speech is learned.
Research on adult speech production suggests that we continue to access and process speech at the syllable level rather than assembling every word sound-by-sound each time we speak (Cholin et al., 2006; Schiller et al., 1996).
When you put that together with what we know about babbling, phonological development, speech motor control and now even artificial speech learning, a consistent picture starts to emerge. The syllable, and particularly the consonant-vowel movement sequence, seems to be doing far more work than many of us were taught to appreciate.
Different Syllable Shapes Need Different Solutions
One reason I dislike blanket statements such as "never segment" is because different word shapes create different motor challenges.
Different syllable structures often require different problem-solving approaches.
For CV words, I might focus on helping the child establish the movement.
For VC words, I may briefly separate the final consonant to help the child become aware that it exists.
For CVC words, I generally preserve the CV movement:
pu - t
rather than:
p - u - t
For CVCV words, I often use natural syllable breaks:
noo - dle
The key point is that I'm not randomly chopping words apart.
I'm trying to identify the movement that is preventing success and making that movement easier to access.
Stress Matters in Longer Words
When working on three-syllable words, I don't just think about syllables.
I also think about stress.
Stress isn't simply a linguistic feature.
It's part of the motor plan.
For example:
to-MATO
rather than:
to-ma-to
Or:
ANNA-mal
rather than:
an-i-mal
I join the stressed syllables together and leave the unstressed syllable on its own.
Why?
Because not all syllables contribute equally to the motor pattern.
The stressed portion of a multisyllabic word often contains the strongest and most stable movement sequence. Preserving that movement pattern can make the entire word easier for the child to access.
"But Won't I Be Teaching the Wrong Motor Plan?"
This is probably the most common question speech therapists ask me.
And it's a reasonable concern.
If I teach:
pu - ppy
instead of puppy
am I teaching the wrong motor plan?
Won't the child become stuck saying it that way forever?
My answer is always the same:
I've never seen that happen.
Not once.
In more than 20 years of using this approach, I have never had a child get permanently stuck at the segmented level.
Why?
Because segmentation is not the goal.
It's a temporary scaffold.
As soon as the child can access the movement, we immediately begin working towards smoother, more connected productions.
Children naturally move toward efficiency. They want the easier, faster motor plan.The risk isn't that the child will get stuck with a temporary scaffold. The bigger risk is that we leave them practising an unsuccessful motor plan.
Children Practise Words Outside Therapy
This is the part I think clinicians sometimes forget.
Children don't only practise words during therapy.
They practise them all day.
Every day.
If a child cannot produce a word and continues repeating it unsuccessfully, they are still rehearsing a motor plan. It's just not a particularly helpful one.
If we can provide a more accurate approximation of the movement, even if it isn't completely smooth yet, we may be giving the child a better pathway toward the final target.
It doesn't have to be perfect.
It just has to be better.
So Should We Never Segment?
I don't think that's the right question.
The reality is that many children with CAS are already segmenting words themselves.
Even when our goal is to maintain the whole word and support smooth coarticulatory transitions, we often see children slowing down, pausing and separating parts of words as they struggle to coordinate the movements.
If you've watched Dr. Edythe Strand's therapy videos, you may have noticed children doing exactly this as they attempt increasingly complex words. It's not necessarily because a previous speech pathologist taught them to segment. Sometimes they're simply trying to find a way to put the word together.
So if a child is already breaking a word apart, why not help them do it in a way that preserves the movement patterns we're ultimately trying to teach?
Instead of separating the initial consonant from its vowel:
The aim is not to teach the child to speak in separated pieces. It's to provide an achievable starting point while preserving as much of the intended word's syllabic structure, movement patterns and stress as possible.
As the child gains control, we gradually reduce the pause and bring the word back together. Sometimes that happens within the same session. For other children, it may take weeks of carefully supported practice. We can't force the word to come together before the child is ready.
And if slowing down, elongation, simultaneous production or chaining already allows the child to produce the word successfully, we may not need to introduce segmentation at all.
Perhaps the issue was never simply whether children with CAS segment. Perhaps it's when, where and how we support that segmentation — and, just as importantly, how we help them bring those movements back together.
Want to go deeper?
Understanding when, where and how to simplify a word is one of the foundations of my approach to severe CAS treatment.
Inside CAS Academy, I go much deeper into:
Analysing syllable shapes
Deciding when to preserve CV movement units
Selecting treatment targets
Adjusting complexity as a child progresses
Using approximations strategically without creating dependency
Making real-time clinical decisions when a child gets stuck
You'll also be able to watch extended real assessment and therapy footage so you can see exactly how these decisions play out with real children in real sessions, not just read about them in theory.
Because knowing that a child is struggling is only the beginning.
The real skill lies in knowing how to make the movement achievable while still preserving the motor plan you're ultimately trying to teach.
References
Jacks, A., Marquardt, T. P., & Davis, B. L. (2006). Consonant and syllable structure patterns in childhood apraxia of speech: Developmental change in three children. Journal of Communication Disorders, 39(6), 424–441.
Cholin, J., Levelt, W. J. M., & Schiller, N. O. (2006). Effects of syllable frequency in speech production. Cognition, 99(2), 205-235.
Jusczyk, P. W. (1997). The Discovery of Spoken Language. MIT Press.
MacNeilage, P. F., & Davis, B. L. (2000). On the origin of internal structure of word forms. Science, 288(5465), 527-531.
Schiller, N. O., Meyer, A. S., Baayen, R. H., & Levelt, W. J. M. (1996). A comparison of lexeme and speech syllables in Dutch. Journal of Quantitative Linguistics, 3(1), 8-28.
Vihman, M. M. (2019). Phonological Templates in Development. Oxford University Press.
Xu, Y. (2025). Syllable as a synchronization mechanism that makes human speech possible. Brain Sciences, 15, 33.
Comments