20Q: Applying Complexity Research to Promote Rapid Improvement for Children Who are Highly Unintelligible

From the Desk of Ann Kummer

Pediatric speech-language pathologists (SLPs) often have children on their caseloads who present with a severe speech disorder, causing their speech to be highly unintelligible. To be most effective, it is important that we target misarticulated phonemes in an appropriate sequence and with appropriate strategies in order to improve speech intelligibility in the shortest period of time. So, how can we determine the appropriate sequence of phoneme selection and also choose effective strategies for rapid progress? Well, this question (and many others) will be answered in this 20Q article by Terri Farnham, who is an expert on this topic.

Teresa (Terri) Farnham, MA, CCC-SLP is a graduate of Case Western Reserve University. She has been practicing speech-language pathology for more than 40 years in a variety of settings, including 20+ years in schools. She is currently in private practice with Clarity for Communication LLC. She has presented around the country on professional issues, speech sound disorders, and augmentative and alternative communication. She is a past President of OSSPEAC (Ohio School Speech Pathology Educational Audiology Coalition) and has also served OSSPEAC as the SLP-At-Large, Conference Chair, and currently as the Government Advisory Committee (GAC) Representative. Terri was the 2018 recipient of the Patricia Lindamood Clinical Leadership Award for her work in phonology therapy.

In this 20Q article, you will learn how to identify characteristics of a child’s phonological acquisition and how this should influence your target selection. You will also learn how to use phoneme contrasts and other key strategies for effective phonological intervention.

Now…read on, learn, and enjoy!

Ann W. Kummer, PhD, CCC-SLP, FASHA, 2017 ASHA Honors
Contributing Editor 

Browse the complete collection of 20Q with Ann Kummer CEU articles at www.speechpathology.com/20Q

20Q: Applying Complexity Research to Promote
Rapid Improvement for Children Who are Highly Unintelligible

Teresa Farnham

1. Could you describe the complexity approach for speech sound disorders? How is it different from traditional speech sound therapy?

I first began learning about the research of Judith Gierut and others in the 1990s. Researchers were talking particularly about targeting later developing, more complex sounds first rather than the usual practice of starting treatment with early developing sounds. I was working in a preschool program at the time with some children with significant intelligibility issues, so I started targeting later developing targets with some of them, with good success. It was surprising how they were able to imitate later developing sounds, even when they didn’t have early developing sounds.

Further reading showed that there were additional considerations for target selection. Gierut’s research pointed to selecting targets that were not only later developing and more marked, but also phonemes that were completely absent from the child’s system and that were not stimulable. So, in a nutshell, the complexity approach directs us to choose targets that our clients know the least about. Such phonemes, for a given child, are considered the most “learnable.”

2. What do you mean by “learnable”?

I like to explain learnability from a language acquisition perspective: since I am a native speaker of English, having spoken it for longer than I care to admit, English is no longer learnable for me; I have already learned it. In contrast, any Southeast Asian language would be highly learnable for me: I have absolutely no knowledge of any of such language’s orthography, tonality, phonology, syllable structure, syntax, pragmatics or semantics. I could decide to try to learn Thai via total immersion in a Thai preschool, where the language would be simpler, but that experience would not provide me with the language input needed for me to truly learn the language: I would not have heard complete, adult models. I would be modeling “toddler Thai”: phrases equivalent to “Mine!”  or “Me go now.”  I would receive a very incomplete exposure to the language system I was trying to learn. While initially simpler to imitate, incomplete language models would actually impede my language acquisition. Total immersion with adult language would produce better language learning.

I encourage you to read Gierut’s 2004 article on learnability as it provides a thorough explanation of the concept as applied to phoneme acquisition. Targeting more complex, unknown sounds works much the same way as learning a new language: unknown, complex phonemes are very “learnable” and provide the child with a more complete picture of the speech sound system. The child begins to derive specific phonologic knowledge from that picture of “the whole.” (Wait! Isn’t that what children with “typically developing” speech do? Hmmmm…)

3.That still seems really counterintuitive. What about being “developmentally appropriate”?

Indeed, those are the magic words in a preschool setting. I too was concerned about trying to do tasks that would be too difficult: especially that the child would be discouraged or resist participating.  I’m happy to report my fears were not realized. In fact, in a study comparing traditional therapy with use of more difficult targets, Rvachew and Nowak (2001) found that both the traditional therapy group and the complex targets group expressed equivalent enjoyment of therapy.

If you’ve ever watched infants learning to walk, you know that they love the challenge: falls do not deter them from getting up again. I have found the same to be true for children learning sounds that are the most difficult for them to produce - they just keep enthusiastically trying! In fact, I think this approach is truly “developmentally appropriate” because it builds on the natural learning patterns and curiosity of children, and presents speech sound information in a more naturalistic way.

One more important consideration for identifying “developmentally appropriate” phonemes: current research in speech sound acquisition. Crowe and McLeod (2020) looked at English speech sound acquisition in the United States by reviewing studies 1931 - 2019. Their conclusions: 90% of children have acquired all plosives, nasals and glides by age 3-11, they add /v, s, z, ʃ, l, ʧ, ʤ/ by one year later, and finally /ð, ʒ, ɹ, θ/ by age 6-11.

A thought worth considering: if all children are acquiring all the consonants of English throughout their toddler and preschool years, perhaps the teaching of any consonant is “developmentally appropriate.”

4. What else did Gierut’s research say about selecting target phonemes for therapy?

Gierut studied target selection choices to see how to make speech sound therapy more efficient. She found the following types of targets produce the greatest system-wide change:

• Later developing phonemes
• Phonemes that are absent from the child’s system
• Phonemes that are not stimulable
• Complex phonemes, possibly clusters
• 2 targets that are absent from the child’s system, and that differ from each other in place, voicing, manner, and major sound class (sonorant/obstruent)

Targeting errors that meet these criteria resulted in the addition of untreated sounds, widespread improvement across sound classes, and reduced over-generalization of targets.

5.    How do children respond to therapy that pushes the limits like this seems to?

Being hesitant to choose /r/ as a target for young children when I began trying this therapy, I often opted for absent phonemes /sh/ vs, /g/:  /ʃ/ is late developing and /g/ is early developing; they are maximally different from each other, except for major sound class. This contrast and similar near-maximal contrasts consistently resulted in system-wide change for most children. My trepidation about /r/ is over; now I frequently use /r/ as an initial target. With this approach, children consistently achieve conversational mastery of nearly all consonants during the space of one school year. Because progress happens so quickly, the children are very aware of their gains, and develop intrinsic motivation to “do hard stuff.”

6.    I’ve always selected target phonemes based on stimulability, consistency, or phonological patterns (like stopping or backing).  Since these kids have so many error sounds, what process do you use to select the most “learnable” target phonemes?

Generally, the highly unintelligible child demonstrates several non-stimulable target possibilities, so limiting target choices can be difficult. Over time, I developed a 5-step process to assist in narrowing the range of error phonemes to the best choices:

• After administering and scoring an articulation test, list all the consonants that the child did not produce correctly in any context. These sounds represent the child’s LEAST phonological knowledge.
• Test all the consonants from step 1 for stimulability, identifying non-stimulable sounds.
• From the non-stimulable phonemes, select the phonemes that are typically acquired later.
• From the later acquired phonemes in step 3, choose the most complex phonemes. This may include consonant clusters.
• From the phonemes remaining on the list, choose two targets that are maximally different from each other. If possible, they should differ in place, voice, manner and major sound class (sonorant/obstruent).

7.    Should I choose two targets that are ABSENT from the child's system to contrast, or should I choose one that they know and one that is absent?

While multiple contrasts, including contrasting unknown vs, known sounds, is an acceptable option, the most efficient option is to choose 2 unknown/absent sounds to use in contrast with each other (Gierut, 1990). The most wide-spread learning occurs when 2 unknown phonemes are treated. This makes therapy a bit more challenging in the beginning because the child has so much to learn about each target phoneme, but the results can be dramatic.

In the initial phase of therapy, my approach is to select one feature of each target sound (e.g., “tongue behind your teeth” or “lips round”) and reward compliance with that one feature, gradually adding more features. When attempting to imitate a maximal contrast pair, if the child produces only one feature correctly, the words will sound different because that one feature is, by definition, different from any of the features of the maximally different opposing target. It is possible to generate lots of incremental success this way, even when the targets are very difficult.

8. How do you go about identifying 2 targets?

First, the targets should be maximally different from each other. We have all worked on reducing error processes by using contrast pairs, usually minimal pairs which contrast the child’s error with the target sound, such as tap/cap for fronting or tame/same for stopping. The complexity model, on the other hand, uses maximal contrasts to provide the child with as much information as possible about the sound system, which promotes system-wide change. Ideally, the targets should differ in all four phonetic parameters: place, voice, manner and major sound class (sonorant/obstruent). Some examples of maximally contrasting consonants are:

• /r/ vs, /s/ - palatal voiced liquid sonorant vs, alveolar unvoiced fricative obstruent
• /l/ vs, /θ/ - alveolar voiced liquid sonorant vs, interdental unvoiced fricative obstruent
• /r/ vs, /k/ - palatal voiced liquid sonorant vs, velar unvoiced stop obstruent

Of course, we don’t live in an ideal world, and, for a particular child, we may be unable to identify unknown potential targets that differ in all four parameters (place, manner, voicing, and major sound class). Indeed, how many times have you seen a child with a severe phonological disorder who, despite having only 5 or 6 consonants in their sound system, easily produces /r/ and /l/? Since /r/ and /l/ are already in that child’s system, these late-developing sonorants cannot be potential targets. In such cases it is likely that the SLP will have to choose targets that differ by place, voice, and manner, but not by major sound class. 3 out of 4 is not the ideal, but it is ok! Clusters may also be a starting point.

9. Did you say “clusters may also be a starting point”? Really?

Yes, really, although there are a few caveats. Williams (1991) treated clusters in several children in order to test the limits of just how complex the targets could be. What she found was that at least one of two foundational skills had to be in place if the child was to make progress using clusters as targets:

• EITHER the child was at least stimulable for both members of the cluster,
• OR the child was observed to have consonant sequences in their system.

If the child had neither knowledge of the 2 members of the cluster nor consonant sequences in their system, progress was very limited.

My experience agrees with that research: starting from absolute zero to teach clusters is simply too hard. If the child doesn’t meet either criteria above, my approach has been to target singleton members of a cluster (such as /s/ vs, /l/ and then move to clusters when the child is able to produce both members of the cluster fluently in imitated words. I have also learned to listen very carefully for consonant sequences in the child’s spontaneous speech, because they can be easily overlooked among the child’s numerous errors. For example, if a child says /pweɪ/ for “play,” that child has demonstrated the ability to produce a consonant sequence, /pw/, even if it is incorrect. In such a case, it is possible, even helpful, to begin with clusters.

Recently, I have been working with a child who is highly unintelligible, with only 24% of words produced completely correctly in a language sample. However, he does have /r/ and consonant sequences such as /pweɪ/ in his system, so /sl/ vs, /gr/ was selected as the target. He already had both /r, l/ in his system, but /s/ and /g/ were absent and not stimulable. During a reassessment of spontaneous word production after 14 treatment sessions with this contrast, percent of words correct in spontaneous speech had risen to 66%, although his production of targets was still inconsistent in imitation.

10. Do you use traditional methods for targeting deletions, or incorporate maximal contrasts somehow?

Deleting a phoneme is the most serious phonemic error a child can make, so deletions should be addressed directly early and often. The child I just mentioned deletes nearly all consonants in the final position. Since the target clusters /sl/ vs, /gr/ don’t occur in the final position, different maximal contrasts had to be selected for final position instruction.

For the clusters above, /s/ and /g/ are the contrasting unknown phonemes, so /z/ vs, /k/, the voicing cognates of /s/ and /g/, were chosen to contrast in the final position. During therapy, the words in a contrast pair (e.g., bees/beak) have had to be produced in segments, “/bi/-/z/“, in order for the child to be able to shape and insert the unknown nonstimulable target in the deleted position. Again, I began with teaching one feature for each target. Gradually, over a number of sessions, he shaped the approximations into the correct sound. He is now able to imitate those final targets fluently and is using an increasing number of final consonants in conversation. Be patient: while those two starting targets are slowly being developed, many other phonemes will be falling into place with very little direct intervention.

While it is absolutely essential to work on the unknown targets in the deleted position, keep in mind that you have selected targets that are not produced in ANY position correctly. They will need to be addressed both in the initial position and any deleted position. Initial contrast words are essential, as the phoneme in the initial position is much more salient for the child. Teaching all positions heightens the child’s awareness of the sound, and facilitates acquisition of the targets.

The challenge for the SLP working on final consonant deletions is finding maximal contrast pairs for the targets. Initial contrasts are easy - just start rhyming! For final contrasts, the simplest solution is to open a speech sound word-list book, find the final position list for one of the targets you have selected, and work your way through that list, substituting the opposing target in the final position to see what contrast words you can create. For example, if I were contrasting /l/ and /ʧ/, I would look up /ʧ/ final words, and substitute /l/ for /ʧ/. Here’s a shortlist, with /l/ substituted for /ʧ/ and a check if the result is an actual word. You should try to have at least 12 pairs to work with.

11. How long would you continue with the original targets before deciding to try something different?

As you work on difficult targets, the child is gaining knowledge about all aspects of speech sounds: place, manner, voicing and sonorance/obstruence. They begin to independently apply this knowledge to less complex phonemes without direct intervention.  In fact, untreated sounds often emerge while the first targets are still unchanged.

These initial targets often take at least 8-10 weeks just to be imitated correctly, so don’t give up. I continue working on the same targets until the child is able to imitate all the contrast pairs fluently, meaning able to repeat the contrast pair accurately on the first try, without any segmentation or obvious effort. This will take what seems like a very long time. For example, one of my kindergarten students had been working on /s/ vs, /r/ as targets from September through January, yet was still struggling each time to blend the fricative /s/ into the vowel that followed it. Then, one day, she succeeded! 2 or 3 sessions later, she had mastered /s/ and all other fricatives in conversation, though only /s/ was treated directly. Her intelligibility changed almost overnight. A long wait and very hard work, but worth it!

12. Just to clarify: when working with a student, do you only choose one maximal contrast pair (which may include voicing cognates) for the whole school year?

To start, I have chosen a pair of unknown phonemes that have maximal contrast with each other. Generally, these first phoneme targets remain the targets for the first 12-15 weeks of therapy. When the child is able to imitate the first contrasts easily and consistently correctly in word pairs, I review the child’s remaining errors and select two more maximally different phonemes. We continue with the new contrast until the child again easily imitates all the new word pairs. Typically, since the first two targets are so difficult, and therapy is a whole new experience for the child, it will take 8-10 weeks before the child even begins to be successful repeating the initial contrast word pairs. Even though the actual targets may seem to be frozen, improvement is apparent for many less complex sounds in conversation during this time. Have patience and keep your ears open!

Once the child is able to imitate the contrast pairs fluently, I move on to either clusters or another maximal contrast pair. Sometimes, only one of the 2 targets is fluent in imitation. In that case, I look for a different maximal contrast for the remaining target, and develop a new round of contrast pairs. The surprising thing is that it is not unusual to have rapid improvement in intelligibility while you’re teaching the first contrast, even though the actual targets do not appear to be improving.

13. What groups of children benefit from this approach? I’m thinking a child with an intellectual disability might not respond well to this level of difficulty.

I have used this approach with great success with children of all educational levels who are highly unintelligible, including children with significant intellectual disabilities as well. Here’s what I think happens:

When minimal contrasts are the focus, the child must produce one and only one feature correctly to make the contrast work (e.g., place for tan/can). This is a tiny target. On the other hand, when we target maximally different phonemes, we are making the child’s overall target quite large. For example, I have often used /ʃ/ vs, /g/ as a contrast. /ʃ/and /g/ differ on 3 of the 4 parameters - place, voice, and manner: /ʃ/ is a noisy (though voiceless) continuant sound with all the energy focused at the front of the mouth; /g/ is a voiced stop, focused at the back of the oral cavity. When producing a pair such as show/go, the child can easily achieve rounded lips and generate frication, even if the /ʃ/ is not quite precise. The contrast with /g/ is huge and allows the child to move their tongue posteriorly in the oral cavity quite readily. The child may make lingual contact at the palate rather than the velum at first, but even then, the distinction between show/go is audible. (That palatal stop sounds a lot more like /g/ than an alveolar /d/ does!) The enlarged target makes it much easier for the child with intellectual or motor challenges to come close to the target sound. Their substitution error has been entirely removed from the process, reducing potential confusion. They may not hit the perfect consonant bullseye, but they can hit the white outer rim of the target, and gradually move their production of the sound toward the center.

14. Going back to the more typical child who would be participating in this type of therapy: how much do I take into account this child’s age when choosing a maximal contrast? If the child is 4 years old, do I choose a contrast that Crowe and McLeod list at age 4 and below?  I would like to contrast /f/ vs, /l/ for a particular child. What would you choose?

First, I think the Crowe & McLeod (2020) information serves to reassure us that selecting the most difficult targets is not really so far from “typical.” Every English-speaking child is learning all the sounds of English from the very beginning. So target the most difficult, latest developing sounds, even if those targets are fully acquired at an age beyond the child’s chronological age.

That being said, later developing sounds make the most efficient targets, so we should also consult Crowe & McLeod to find out age of acquisition for all the child’s errors. Their work indicates that /f/ is an early developing fricative, so it would be a less efficacious target. Instead, I would choose a later-developing affricate, such as /ʧ/, because by teaching /ʧ/, /f/ and all other fricatives should appear in the child’s system without direct intervention.  This type of generalization doesn’t occur in the other direction: teaching /f/ will not generalize to /ʧ/.  /tʃ/ vs. /l/ could be a great contrast choice, if /tʃ/ is one of the child’s nonstimulable errors.

Keeping least phonological knowledge, age of acquisition, and stimulability in balance is a bit of a juggling act. That is why I developed the 5-step sifting process seen in question #4.

15. What does therapy look like? Are there ways to use this approach to also support literacy development?

I typically see children individually for 30 minutes once a week. This is sufficient to move most children from highly unintelligible speech to demonstrating few or no speech sound errors during the course of a school year. Each session has 4 key parts which also support literacy development. The four parts are:

1. Stimulability and letter-sound correspondence practice: this activity simply uses print representations of all the consonants in order from front to back, posted in order around the speech mirror. The child “leads” the activity by using a magic wand to point to each letter and say or imitate the given phoneme. Stimulable sounds are much more likely to get into the child’s system without direct intervention, so this activity is helping speed up the child’s progress. It also gives the child lots of practice with letter-sound correspondence, which they may have missed in the regular classroom because they couldn’t be understood. A chart of the letters in order from front to back is available free at my TeachersPayTeachers page (ClaritySLP).
2. Phoneme segmentation and blending with maximal contrast pairs: Phonologic and phonemic awareness are essential to developing literacy. I have modified an activity called “Say-It Move-It” from Road to the Code (Blackman et al, 2000) to assist in learning phoneme segmentation and blending. The activity requires sliding a chip or marker to a line as you and the child say each sound in a word in unison, one at a time - “/ʃ/ - /o/” - and then blend the sounds back together - “show!” Using the contrast pairs for segmentation practice allows the child to practice their targets in isolation, yet the target is still in the context of a real word. After the first word in the pair is segmented and blended, we remove the contrast consonant marker (in this case, /ʃ/) and place a different-colored marker in its place for /g/. We say “/g/ - /o/” in segments while tapping each marker, and then blend it back into “go.” My TeachersPayTeachers page (ClaritySLP) also has complex maximal contrast pairs for practice and use in a game format.
3. Pair practice for fluency in a game setting: this gives the child an opportunity to practice the maximal contrast pairs as whole words.
4. Practice target phoneme in high frequency words: Morrisette & Gierut (2002) found that teaching the target sound in high frequency words facilitates change in the child’s overall system. I select one high frequency word for each target. For our example targets, /ʃ/ and /g/, “she” and forms of “go” are great high frequency words for practice. These are words that are flexible enough to use in a variety of sentences. “She is going for a bike ride.” “She wants to go to the store.” “Can she go too?”

Each of these four activities addresses early literacy skills: letter recognition, letter-sound correspondence, onset identification, phoneme segmentation and blending, formulating syntactically correct sentences. These skills are taught early in the classroom, but by the time the child with a phonological disorder is ready to learn these skills and effectively demonstrate competence, the rest of the class may have moved on. Addressing these skills with the child consistently and systematically during therapy is vital to their acquisition of early literacy skills.

16. When choosing targets, is it okay to choose an earlier developing sound to compare with a more complex sound if it has a greater contrast?  Otherwise, I feel like we will always end up with /θ/ and /r/.

/θ/ vs. /r/ is actually a very powerful contrast that helps elicit /r/ quite effectively, but you’re right: teaching the same contrast to everyone because they are the last to be acquired is not how to approach individual children. Early developing phonemes such as /k, g/ are commonly in error for young children and often not stimulable. If /k, g/ are absent from the child’s system and not stimulable, they, as a cognate pair, could be a single efficacious target. /k, g/ might also provide a better phonetic contrast to another unknown, but later-developing target. For example, /ʧ/ and /r/ seem like good targets because they are both later developing sounds, and complex. /ʧ/ and /r/ differ in voice, manner and major sound class, but not place. They are both palatal sounds. In actuality, /ʧ/ vs. /r/ is not a successful contrast because it is very difficult for the child to put their tongue in the same place, and then produce 2 different phonemes.  Place is an important phonetic distinction for young children. So, if the child has no phonologic knowledge of /k, g/, those sounds would be a better contrast to /r/ despite their early acquisition. I would also note that /k, g/ are later-developing stops, and so might facilitate improvement of inconsistent /p, b, t, d/, which are earlier acquired stops.

17. If the maximal contrast pairs you have selected contain other difficult sounds (e.g., the "r" in “real" when targeting final /l/), do you address those other errors during a session, or would you just overlook an error for /r/ and concentrate on the target /l/?

I usually provide a little coaching on non-target error sounds that may occur in the target pairs, but only for 2 or 3 trials before moving on. Those sounds were addressed during the stimulability activity, and they will continue to improve in that context. Even during the stimulability activity, errors are only coached very briefly, for the purpose of shaping correct production over time.

18. If you are working on both initial and final consonant deletion, would you focus on initial consonants one week and then final consonants the next week? Or would you focus on both in the same session?

For the phoneme segmentation and blending task, I tend to start with initial contrasts because initial consonants are more salient to young children. After the child has the basic idea of phoneme segmentation and blending in imitation, I quickly add final consonant contrasts to the activity, and use both during all sessions going forward. If the child continues to have greater difficulty with final consonants down the road, I increase therapy time spent on final consonant contrasts.

19. When teaching clusters are you choosing a variety of clusters? What kind of contrasts do you use for clusters?

If the child is a candidate for clusters (see question 8), there are two possibilities for contrasting the clusters: 1) contrasting the target cluster with another cluster (as in slip/grip) or 2) contrasting the cluster with an unknown singleton (slob/job). I usually try to avoid narrower cluster reduction contrasts (such as slip/sip/lip) because the contrast is minimal. The larger the contrast between the two targets, the more likely the child is to hit the target.

20. Do you have any final advice for clinicians who want to get started treating complex, “learnable” sounds?

Be patient. Teaching difficult things is, well, difficult. Be prepared to apply everything you learned about phonetics in grad school to building a scaffold for your clients to acquire difficult sounds. And then, stand back and watch the progress begin.

This is a challenging approach to selecting and treating phonological targets, but extremely rewarding. Thanks for joining the conversation!

References

Baker, E., and McLeod, S. (2011). Evidence-Based Practice for Children with Speech Sound Disorders: Part 2 Application to Clinical Practice. Language, Speech and Hearing Services in Schools, 42; 140-151.

Blachman B.A., Ball E.W., Black R., Tangel D.M. Road to the Code: A Phonological Awareness Program for Young Children. Baltimore, MD: Paul H. Brookes Publishing Company; 2000.

Creaghead, N., and Farnham, T. (2013). Accelerating Progress for Young Children with Speech Sound Disorders. Short Course, American Speech-Language-Hearing Association Convention.

Crowe, K, and McLeod, S. (2020). Children’s English consonant acquisition in the United States: a review. American Journal of Speech-Language Pathology, 29, 2155 -2169.

Farnham T. (2014). Articulation and Phonology Guidelines. http://omnie.ocali.org/mod_view.php?nav_id=268.

Gierut, J. A. (1989). Maximal Opposition Approach to Phonological Treatment. Journal of Speech and Hearing Disorders, 54(1), 9-19.

Gierut, J. A. (1990). Differential Learning of Phonological Oppositions. Journal of Speech and Hearing Research, 33(3), 540-549.

Gierut J.A. (1992). The conditions and course of clinically induced phonological change. Journal of Speech and Hearing Research. 5(5):1049-1063.

Gierut J.A. (1998). Treatment efficacy: functional phonological disorders in children. Journal of Speech and Hearing Research. 41(1):S85-S100.

Gierut, J.A. (2001). Complexity in Phonological Treatment: Clinical Factors. Language, Speech and Hearing Services in Schools, 32, 229 – 241.

Gierut, J.A. (2004). The Learnability Project. The ASHA Leader, 9(22): 28.

Gierut, J.A. (2007). Phonological Complexity and Language Learnability. American Journal of Speech-Language Pathology, 16, 6-17.

Gierut J.A., Elbert M., Dinnsen D. (1987) A functional analysis of phonological knowledge and generalization learning in misarticulating children. Journal of Speech Language and Hearing Research. 30(4):462-479.

Gierut J.A. Morrisette M.L., Hughes M.T., Rowland S. (1996).  Phonological treatment efficacy and developmental norms. Language, Speech and Hearing Services in Schools. 27:215-230.

Gierut, J.A., Morrisette, M.L., Hughes, M.T., & Rowland, S. (1996). Phonological treatment efficacy and developmental norms. Language, Speech and Hearing Services in Schools 27;215-230.

Gierut, J.A. Morrisette, M.L., Ziemer, S.M. (2010). Nonwords and generalization in children with phonological disorders. American Journal of Speech-Language Pathology. 19:167-177.

McLeod, S., & Crow, C. (2018). Children’s consonant acquisition in 27 languages: a cross-linguistic review. American Journal of Speech-Language Pathology, 27, 1-26.

Morrisette M.L., & Gierut, J.A (2002). Lexical organization and phonological change in treatment. Journal of Speech Language and Hearing Research, 45(1):143-159.

Powell, T. W. & Elbert, M. (1984). Generalization Following the Remediation of Early- and Later-Developing Consonant Clusters. Journal of Speech and Hearing Disorders, 49(2), 211-218.

Powell T.W., Elbert M., & Dinnsen, D.A. (1991). Stimulability as a factor in the phonological generalization of misarticulating preschool children. Journal of Speech and Hearing Research, 34(6):1318-1328.

Rvachew, S., & Nowak, M. (2001). The effect of target selection strategy on phonological learning. Journal of Speech Language and Hearing Research, 44(3):610-623.

Storkel, H. (2019). The complexity approach to phonological treatment: how to select treatment targets. Language, Speech and Hearing Services in Schools, 49, 463-481.

Williams A.L. (1991). Generalization patterns associated with training least phonological knowledge. Journal of Speech and Hearing Research, 34:722-733.

Williams A.L. (2000). Multiple oppositions: theoretical framework for an alternative contrastive intervention approach. American Journal of Speech-Language Pathology, 9:282-288.

Williams A.L. (2005). Assessment, target selection and intervention: dynamic interactions within a systemic perspective. Topics in Language Disorders, 25(3):231-242.