Research has demonstrated that it is most efficacious to treat three-element clusters to effect change throughout a child's system (if the child has the second and third consonants in his/her system already). For example, if a child is taught /spl-/, then he/she should already have /p/ and /l/ in his/her phonemic repertoire. If the child has a limited phonemic inventory, this would not work.
However, there has been some exciting research that would be applicable to these students and others who have difficulty with clusters. The Sonority Sequencing Principle for Clusters was identified in linguistic research. It dictates that onsets (word-initial sounds) must rise in sonority and codas (ending sounds) must fall in sonority. Sonority is the inherent loudness of sounds relative to one another. The greater the sonority, the wider the mouth is and the more vowel-like a sound is (Barlow, 2000). Linguists have identified the relative sonority for different sound classes. In 1990, Steraide assigned relative values to each sound class, indicated in parentheses. The most sonorous sounds are vowels (0), followed by glides (1), liquids (2), nasals (3), voiced fricatives (4), voiceless fricatives (5), voiced stops (6) and voiceless stops (7).
In 1999, Gierut applied this principle to treatment. She identified an implicational relationship between clusters with small sonority distances and clusters with large sonority distances. If a child's system included clusters with a small sonority distance, that implied the presence of clusters with a large sonority distance. For example, the cluster /sn-/ has a small sonority distance (voiceless fricative Steraide's value of 4) - nasal (value of 2)) of 2. The cluster /kw-/ has a large distance of 6 (voiceless stop (value of 7) - glide (value of 1)). Therefore, the cluster /sn-/ is considered to be more complex and marked relative to /kw-/.
This resource displays a hierarchy amongst two-element clusters that demonstrates that some two-element clusters are more complex than others and that these should be targeted to create the most change. (Three-element clusters abide by different rules.) The attached document (Sonority Sequencing Principle for Clusters) identifies the sonority distance between clusters. Research has suggested that it is most efficacious to teach clusters with a small sonority difference, such as /fl-/, /fr-/, etc., to create change in a child's system. One caveat is to avoid teaching /sp-/, /sk-/ and /st-/ because they violate the Sonority Sequencing Principle, resulting in a sonority difference of -2. Furthermore, Gierut and her colleagues found that teaching these three fricative + stop clusters inhibited generalization to other clusters. Linguists do not consider /sp-/, /sk-/ and /st-/ to be true clusters. /sn-/ and /sm-/ also often pattern in the same way and should not be targeted in those cases.
Please see the attached document, Sonority Sequencing Principle for Clusters for the values of each cluster. If SLPs teach the most complex clusters, children acquire the less complex clusters without direct treatment.
Interested readers are directed to Gierut’s 1999 paper entitled Syllable Onsets: Clusters and Adjuncts in Acquisition that illustrates these principles. The paper can be downloaded here.