Manipulating Linguistic Complexity to Improve Child Language Treatment Outcomes

Phonological disorder (PD) and specific language impairment (SLI) directly impact a child's ability to communicate and are among the most prevalent developmental disorders. The proposed experiments manipulate the complexity of treatment targets to identify the most efficacious treatment approaches for English- and Spanish-speaking children aged 3 to 6 years who present with these disorders. This research will reveal the nature of interactions between sound and structure in language for these children and will have significant implications for a unique approach to target selection when treating persistent phonological and grammatical difficulties in children with PD, SLI, or both.

Specific Aim 1: To evaluate relative effects and effectiveness of phonological complexity of treatment stimuli on generalization learning following speech-sound treatment of Spanish-speaking children with PD. Experiment 1 is an across-subjects evaluation of phonological complexity of treatment words on speech-sound generalization learning in Spanish-speaking children with PD. Children will receive speech-sound treatment in Spanish on words that are phonologically simple or complex. It is predicted that the phonologically complex cluster condition will lead to greater generalization as compared to the singleton condition in the children's speech sound use. Specifically, it is assumed that greater generalization learning will occur following treatment on consonant clusters than following treatment on singletons. These predictions will be evaluated based on the children's posttreatment accuracy on the Generalization Probe. Specific Aim 2: To evaluate relative effects and effectiveness of phonological and morphological complexity of treatment verb stimuli on generalization learning following speech-sound treatment of English-speaking children with PD and PD-SLI. Experiment 2 is an across-subjects evaluation of phonological and morphological complexity of treatment words on speech-sound generalization learning in children with PD and those with PD-SLI. Children will receive speech-sound treatment in English on verbs that are morphologically simple or complex in phonologically simple or complex verbs. It is predicted that the phonologically complex cluster condition will lead to greater generalization (vs. the singleton condition) in the PD and PD-SLI children's sound systems, and that this effect will be enhanced in a morphologically complex bi-morphemic context. It is assumed that treatment on final clusters will generalize to final singletons and clusters, whereas treatment on final singletons will generalize to final singletons only. Further, it is predicted that treatment in a bi-morphemic context will generalize to mono-morphemic contexts, but not vice versa. Finally, although it is predicted that the cluster advantage will be observed in both populations, it is possible that, for the PD-SLI children, the bi-morphemic context may introduce too much new information (new phonological structure and new morphological structure) that may inhibit generalization learning for those children due to their reduced accuracy in both phonology and morphology. These predictions will be evaluated based on the children's post-treatment accuracy on the Generalization Probe. Specific Aim 3: To evaluate relative effects and effectiveness of phonological complexity of bi-morphemic treatment verb stimuli on generalization learning following morpheme treatment of English-speaking children with SLI and PD-SLI. Experiment 3 is an across-subjects evaluation of phonological complexity of bi-morphemic treatment words on morpheme generalization learning in children with SLI and those with PD-SLI. Children will receive morpheme treatment in English on the third-person singular (3rdSg) morpheme "-s" in phonologically simple or complex verbs. It is predicted that the phonologically complex cluster condition will lead to greater generalization as compared to the singleton condition in the SLI and PD-SLI children's morpheme use. Specifically, it is assumed that greater generalization learning of the treated 3rdSg morpheme will occur following treatment on the morpheme in cluster contexts than following treatment in singleton contexts. Further, though it is predicted that the cluster advantage will be observed in both populations, it is possible that, for the PD-SLI children, the cluster context may introduce too much new information (new morphological structure and new phonological structure) that may inhibit generalization learning for those children due to their reduced accuracy in both phonology and morphology. These predictions will be evaluated based on the children's post-treatment accuracy on the Generalization Probe. This proposed research program will evaluate the influence of morpho-phonological interaction in the language of English and Spanish-speaking children with phonological disorder (PD), those with specific language impairment (SLI), and those children with co-occurring PD and SLI (PD-SLI) through manipulation of phonological and morphological complexity in the selection of treatment target words. The proposed research will 1) bridge linguistic theory with clinical treatment to identify ideal targets for generalization learning across populations, 2) focus on both morphology and phonology, which are known to interact in normal systems, and 3) define how these principles together impact on the management of disordered systems. Upon completion of the proposed work, the nature of morpho-phonological interactions within and across the clinical populations will be identified, and the role of (morpho-)phonological complexity in driving change following treatment will be delineated. The participants for the proposed study will include 10 Spanish-speaking children aged 3 to 6 years with PD, and 36 English monolingual children between the ages of 4 and 6 years of age with PD (n = 12), SLI (n = 6), or co-occurring PD-SLI (n = 18). The participants will be recruited through schools and community organizations but will not be enrolled in any other speech/language services elsewhere. All participants will have typical intellectual, hearing, social-emotional, and neurological development, per caregiver report. Both male and female participants will be recruited, as will children from varied racial/ethnic backgrounds. This study employs a single-subject, staggered multiple-baseline design. Data will include caregiver-reported demographic data, scores on the standardized tests, digital recordings of participants' spontaneous language, Treatment Probes and Generalization Probes, computerized analyses of speech/language samples and probes, and transcribed responses from speech/language samples and probes. All such data will be collected following caregiver consent and child assent. Further, all data will be collected at the San Diego State University (SDSU) Speech-Language Clinic, with the exception of the caregiver report, which may be completed by the caregiver at home. Children who meet the inclusionary criteria will be classified as having PD, SLI, or PD-SLI, per those criteria. They will be assigned to a particular treatment condition. Spanish PD children will be assigned to one of two conditions of Experiment 1 (n= 5 per condition). English PD children will be assigned to one of four conditions of Experiment 2 (n=3 per condition) and SLI children will be assigned to one of two conditions of Experiment 3 (n=3 per condition). PD-SLI children will be assigned to one of four conditions of Experiment 2 (n=3 per condition) or one of two conditions of Experiment 3 (n=3 per condition). Following treatment condition assignment, specific treatment target stimuli will be identified for each child. The next set of procedures involve the following, with 1-hour sessions occurring three times weekly: Baseline testing phase, to include administration of the Generalization Probe (a minimum of three baselines per child); Treatment phase (maximum 18 sessions), to include Treatment Probe administered at the start of each session and Generalization Probe collected after 9 treatment sessions; Post-treatment testing phase to include Generalization Probe collected immediately following treatment, and at 2 weeks and 2 months posttreatment. Following treatment condition assignment, specific treatment target stimuli will be identified for each child. The next set of procedures involve the following, with 1-hour sessions two to three times weekly: Baseline testing phase, to include administration of the Generalization Probe (a minimum of three baseline sessions per child); Treatment phase (maximum 18 sessions), to include Treatment Probe administered at the start of each session and Generalization Probe collected after 9 treatment sessions; Post-treatment testing phase, to include Generalization Probe collected immediately following treatment, and at 2 weeks and 2 months posttreatment. Following completion of the posttreatment testing, children will be dismissed from participation in the research program. Data from Treatment and Generalization Probes will reflect measures of the dependent variable. Treatment Probe data will be collected to document learning during treatment. Children's learning of the treated target (i.e., singleton or cluster for Experiments 1 and 2, 3rdSg morpheme for Experiment 3) in treated words across all treatment sessions will be measured. This will be done via measures of percent accuracy on the treatment target in treated words produced during treatment sessions. Generalization data will be collected during pretreatment baseline sessions, after 9 treatment sessions, immediately following treatment, and at 2 weeks and 2 months posttreatment, and will consist of singleton and cluster accuracy and morpheme accuracy on the Generalization Probe. All the above speech/language samples will be orthographically transcribed and coded for morpho-syntax using Systematic Analysis of Language Transcripts (SALT), and phonetically transcribed, using broad notation of the International Phonetic Alphabet in Phon software by trained research assistants who are undergraduate and graduate students of speech-language pathology and/or linguistics. Unintelligible words will not be included in accuracy analyses, given that a target form cannot be determined for such productions. Generalization is operationally defined as greater than 10% increase in accuracy on the generalization data following treatment as compared to average baseline levels. Following completion of the post-treatment testing, children will be dismissed from participation in the research program. Referrals, if needed, for further speech/language services will be provided.

The clinician will provide models, verbal and/or tactile cues, and/or conversational recasts of speech sound targets following the methodology of Gierut and colleagues. Up to 100 productions will be targeted per 1-hour session.

The clinician will provide models, verbal and/or tactile cues, and/or conversational recasts of targeted morphemes following the methodology of Plante and colleagues. A minimum of 50 productions will be targeted per 1-hour session.

The Treatment Probe is made of the selected treatment stimuli (words) that are consistent with the child's assigned experimental condition, and evaluates production accuracy of the treatment target within the verb stimuli. The treatment targets in these words have a consonant singleton or consonant cluster ("tee" vs. "tree", "sees" vs. "seats"), and are mono- or bi-morphemic contexts ("tease" vs. "sees"). Children are asked to pronounce each word following presentation of a corresponding picture and a verbal prompt. The Treatment Probe allows us to track the effectiveness of treatment on the target consonant or consonant cluster in the treated stimuli. Accuracy of the consonant or consonant cluster is binary, as 'incorrect' (e.g., "tee" for target "tree") or 'correct' (e.g., "tree" for target "tree").

Change in percentage of consonant or consonant cluster accuracy (depending on Arm/Group) from first treatment session to final treatment session; up to 6 weeks.

The Generalization Probe consists of words and phrases that target each consonant, cluster, and morpho-syntactic constructs a minimum of 10 times across relevant contexts (i.e., word- and utterance-position). The Generalization Probe allows us to track the effects of treatment (generalization) to untreated stimuli and to monitor control variables by measuring production accuracy of stimuli. The Generalization Probe samples each consonant and consonant cluster of the target language across multiple words and contexts. Children are asked to pronounce each word following presentation of a corresponding picture and a verbal prompt. Accuracy of the consonant or consonant cluster is binary, as 'incorrect' (e.g., "tee" for target "see") or 'correct' (e.g., "see" for target "see").

Change in percent consonant accuracy from baseline to immediately following final treatment session; up to 6 weeks.

Participants will be recruited through announcements to schools and community organizations. Data will include caregiver-reported demographic data, scores on the standardized tests, digital recordings of subjects' spontaneous language, treatment probes and generalization probes, computerized analyses of speech/language samples and probes, and transcribed responses from speech/language samples and probes. All such data will be collected following caregiver consent and child assent. Further, all data will be collected at the SDSU clinic, with the exception of the caregiver report, which may be completed by the caregiver at home. Screening procedures will include administration of the following: Caregiver report of child history and language background, binaural hearing screening, Goldman-Fristoe Test of Articulation 3 (GFTA3; English) or Bilingual English-Spanish Assessment (BESA; Spanish), Preschool Language Scales (PLS; English or Spanish), Leiter-R, spontaneous language sample, and Protocol for the Assessment of English Phonotactics (PEEP; English) or Assessment of Spanish Phonology (ASP; Spanish). Children who meet the inclusionary criteria below will be classified as having PD, SLI, or PD-SLI, per those criteria. INCLUSIONARY CRITERIA: All participating children must meet the following inclusionary criteria: between the ages of 3 and 6 years of age; speak Spanish (Experiment 1) or English (Experiments 2 and 3); present with phonological disorder (PD, n = 22), specific language impairment (SLI, n = 6), or co-occurring PD-SLI (n = 18) (see below for additional criteria for inclusion). The research program will over-recruit at 60 children, to reach the desired 46 participants. Participants will complete a battery of assessment measures. Information gleaned from these measures will be used to further determine eligibility for the proposed study, which include both quantitative and qualitative criteria. Specifically, all PD (Experiments 1 and 2) and PD-SLI (Experiment 2 and 3) participants must: exhibit 5 or more sounds in error across three or more speech sound manner classes; score ≤ 1.5 standard deviations below the mean on the GFTA3, if English-speaking; score ≤ 1.5 standard deviations below the mean on the Phonology subtest of the BESA, if Spanish-speaking; and exhibit ≤ 20% accuracy on consonants and clusters independent of tense morphemes. Moreover, all SLI (Experiment 3) and PD-SLI (Experiments 2 and 3) participants must: score ≤ 1 standard deviation below the mean on the Preschool Language Scales (PLS), a test of expressive and receptive language; and exhibit a mean length of utterance (MLU) ≤ 1 standard deviation below the mean for age- and demographic-matched peers, based on a language sample. EXCLUSIONARY CRITERIA: All participants (Experiments 1, 2, and 3) must: not be receiving speech/language services elsewhere; pass a binaural hearing screening at 20 decibels; achieve score above a standard score of 70 on a test of nonverbal cognition (Leiter-R); and have typical intellectual, hearing, social-emotional, and neurological development, per parent report. In addition, to rule out concomitant difficulty in other domains of language all PD participants (Experiments 1 and 2) must: score > 1 standard deviation below the mean on the PLS, if English-speaking; have an MLU > 1 standard deviation below the mean for age- and demographic-matched peers, based on a language sample. To rule out concomitant difficulty in phonology, the SLI participants (Experiment 3) must: score > 1.5 standard deviations below the mean on the GFTA3; and exhibit fewer than 5 sounds in error and > 20% accuracy on final consonants and clusters independent of tense morphemes. If the above criteria are not met, a child will be excluded from participation.

Combiths PN, Barlow JA, Potapova I, Pruitt-Lord S. Influences of Phonological Context on Tense Marking in Spanish-English Dual Language Learners. J Speech Lang Hear Res. 2017 Aug 16;60(8):2199-2216. doi: 10.1044/2017_JSLHR-L-16-0402.

Pruitt S, Oetting J. Past tense marking by African American English-speaking children reared in poverty. J Speech Lang Hear Res. 2009 Feb;52(1):2-15. doi: 10.1044/1092-4388(2008/07-0176). Epub 2008 Aug 11.

Cummings AE, Barlow JA. A comparison of word lexicality in the treatment of speech sound disorders. Clin Linguist Phon. 2011 Apr;25(4):265-86. doi: 10.3109/02699206.2010.528822. Epub 2010 Dec 15.

Barlow JA. Phonological change and the representation of consonant clusters in Spanish: a case study. Clin Linguist Phon. 2005 Dec;19(8):659-79. doi: 10.1080/02699200412331279794.

Gierut JA. Nexus to Lexis: Phonological Disorders in Children. Semin Speech Lang. 2016 Nov;37(4):280-290. doi: 10.1055/s-0036-1587704. Epub 2016 Oct 4.

Gierut JA, Morrisette ML, Dickinson SL. Effect Size for Single-Subject Design in Phonological Treatment. J Speech Lang Hear Res. 2015 Oct;58(5):1464-81. doi: 10.1044/2015_JSLHR-S-14-0299.

Gierut JA, Morrisette ML. Age of word acquisition effects in treatment of children with phonological delays. Appl Psycholinguist. 2012 Jan 1;33(1):121-144. doi: 10.1017/S0142716411000294.

Gierut JA, Morrisette ML. Effect size in clinical phonology. Clin Linguist Phon. 2011 Nov;25(11-12):975-80. doi: 10.3109/02699206.2011.601392. Epub 2011 Jul 25.

Gierut JA. Phonological complexity and language learnability. Am J Speech Lang Pathol. 2007 Feb;16(1):6-17. doi: 10.1044/1058-0360(2007/003).

Gierut JA. Complexity in Phonological Treatment: Clinical Factors. Lang Speech Hear Serv Sch. 2001 Oct 1;32(4):229-241. doi: 10.1044/0161-1461(2001/021).

Gierut JA. Syllable onsets: clusters and adjuncts in acquisition. J Speech Lang Hear Res. 1999 Jun;42(3):708-26. doi: 10.1044/jslhr.4203.708.

Gierut JA, Champion AH. Syllable onsets II: three-element clusters in phonological treatment. J Speech Lang Hear Res. 2001 Aug;44(4):886-904. doi: 10.1044/1092-4388(2001/071).

Van Horne AJO, Fey M, Curran M. Do the Hard Things First: A Randomized Controlled Trial Testing the Effects of Exemplar Selection on Generalization Following Therapy for Grammatical Morphology. J Speech Lang Hear Res. 2017 Sep 18;60(9):2569-2588. doi: 10.1044/2017_JSLHR-L-17-0001.

Thompson CK, Shapiro LP. Complexity in treatment of syntactic deficits. Am J Speech Lang Pathol. 2007 Feb;16(1):30-42. doi: 10.1044/1058-0360(2007/005).

Thompson CK. Complexity in language learning and treatment. Am J Speech Lang Pathol. 2007 Feb;16(1):3-5. doi: 10.1044/1058-0360(2007/002).

Leonard LB, Deevy P. The Changing View of Input in the Treatment of Children With Grammatical Deficits. Am J Speech Lang Pathol. 2017 Aug 15;26(3):1030-1041. doi: 10.1044/2017_AJSLP-16-0095.

Plante E, Ogilvie T, Vance R, Aguilar JM, Dailey NS, Meyers C, Lieser AM, Burton R. Variability in the language input to children enhances learning in a treatment context. Am J Speech Lang Pathol. 2014 Nov;23(4):530-45. doi: 10.1044/2014_AJSLP-13-0038.

Leonard LB, Davis J, Deevy P. Phonotactic probability and past tense use by children with specific language impairment and their typically developing peers. Clin Linguist Phon. 2007 Oct;21(10):747-58. doi: 10.1080/02699200701495473.

Potapova I, Kelly S, Combiths PN, Pruitt-Lord SL. Evaluating English Morpheme Accuracy, Diversity, and Productivity Measures in Language Samples of Developing Bilinguals. Lang Speech Hear Serv Sch. 2018 Apr 5;49(2):260-276. doi: 10.1044/2017_LSHSS-17-0026.