Fey, Long & Finestack, 2003 1. To help the child achieve greater facility in the comprehension & use of syntax and morphology in the service of conversation,

Intervention for Developing Language

Intervention for the Developing Language Stage

Principles of Grammar Facilitation in InterventionFey, Long & Finestack, 20032From the theoretical perspective that specific language impairment is associated with a general limitation hypothesisPrinciples Related to Goal SelectionTo help the child achieve greater facility in the comprehension & use of syntax and morphology in the service of conversation, narration, and text in both written and oral modalitiesGrammatical form should rarely be the only aspect of language intervention & communication that is targetedStimulate the childs language acquisition processes rather than teach specific language formsSpecific goals should be based on the childs readiness and need for the targeted formsImprovements in grammar will have a positive influence on communication or other areas of developmentContexts arise naturallyWhole languageSuccess is demonstrated by use of targets in meaningful oral or written communication activities1. To help the child achieve greater facility in the comprehension & use of syntax and morphology in the service of conversation, narration, and text in both written and oral modalities

Other linguistic & non-linguistic areas likely require attentionFocus on grammar alone does not necessarily generalize to other areasPhonology, pragmatics, semantics, narration, literacy2. Grammatical form should rarely be the only aspect of language intervention & communication that is targeted

3. Stimulate the childs language acquisition processes rather than teach specific language formsLeads to more efficient and rapid learning throughout their environmentTarget grammatical categories (e.g., all subjective pronouns), principles (e.g., SVO), and operations (interrogative reversal)Use childs existing resources (e.g. take advantage of S-W predispositon)Seek ways in which system-wide change can occur6Strong-weak (trochaic) patterns. Free morphemes are more salient when they follow a stressed syllable than an unstressed syllable: e.g., A dog is big. A tiger is big. I saw a flower. I watered a flower. Jim saw a flower. He saw a flower4. Specific goals should be based on the childs readiness and need for the targeted formsFocus on partially mastered and unused forms for which the child demonstrates cognitive and linguistic readiness and functional needHighest Priority:Forms & functions used with 10% - 50% accuracyHigh Priority:Forms & functions used with 1% - 10% accuracyLower Priority:Forms & functions used with 50% - 90% accuracyForms not understood or used at allTrochaic English is a trochaic language, stress on the first part and not on the last part. BeforeTake advantage of patterns in sentense to help the child.7Principles Related to Intervention Procedures & ActivitiesManipulate the social, physical, and linguistic context to create more frequent opportunitiesExploit different textual genres & written modality to develop appropriate contextsManipulate the discourse so targets are rendered more salient in pragmatically felicitous contextsSystematically contrast the childs forms with the adult form using recastsAvoid telegraphic speech: provide models in well-formed phrases and sentencesUse elicited imitation to make tragets more salients and to provide practice with phonological patterns that are difficult to access or produce5. Manipulate the social, physical, and linguistic context to create more frequent opportunities

Increase opportunities for recasts and modelsViolation of routine eventsWithhold objects and turnsViolate object function or manipulationIntentionally misplace objectsBecome less cooperativeMisname items, actions, or events

Clinic does not have to be inside in a room.9

6. Exploit different textual genres & written modality to develop appropriate contexts

Doing well in conversation with a cooperative partner does not necessarily generalize to other contextsObligatory contexts occur more frequently in some genres and modalities than others (e.g., passives; 3rd person singular)10Past tense and possessives are highly common in narratives but much rarer in conversation3s: not in conversation or narratives, but in expository text. what does a nurse do?7. Manipulate the discourse so targets are rendered more salient in pragmatically felicitous contextsMake targets:Longer, louder, with dynamic pitch changesUse pragmatically appropriate contextsEllipsis in sentence- or phrase-final positionContrast between one assertion and anotherDisagree or teaseSongs

11That boy is really big. Who is? He is.The caterpillar will become a butterfly. I know it will.

Dont put it inside the barn. Put it outside.Lynda wasnt singing, but Joe was.No you cant. Yes I can.

Twinkle twinkle. On the swings.

8. Systematically contrast the childs forms with the adult form using recasts

Clarify the relationship between target form and their semantic/pragmatic/grammatical functionsRecasts:Maintain meaning, but modify (correct) structureProvide additional processing timeFocus attention on the contrastUsed in naturally occurring, pragmatically relevant contexts9. Avoid telegraphic speech: provide models in well-formed phrases and sentencesBecause:Comprehension often > expressionsSensitive to grammatical morphemes in the speech stream before they produce themLess opportunity to hear morphemes in the S-W patternEnglish is a morphologically sparse language, so there are already fewer opportunities to associate meaning to morphemesGrammatical functors are used as cues to grammatical class

10. Use elicited imitation to make targets more salient & to provide practice with phonological patterns that are difficult to access or produce

Use imitation as a mechanism to focus attentionUse imitation to contrast forms and highlight meaningImitation alone is not sufficient to establish learning

Techniques: The Power Tools You Always Have with YouActive Ingredients/Teaching Episodes:The techniques we use to teach or enhance new learning and behavior(Warren, Fey, & Yoder, 2007)

TechniquesTime-delay/slowingModelsRecasts Expansions

MandsQuestions For ImitationDirect InstructionTime Delay/SlowingEnvironmental arrangementWAIT Why Am I Talking ?As long as you canthen count to 5PACEPlay rate Attentional focusCommunication rate Energy levelAlter the auditory characteristics of the inputSlow the rate of inputHighlight targets with slight lengthening and emphasisPause between phrases

ModelsPresentation of language target with or without requiring a child responseUse well-formed phrases & sentencesInclude contrastsIncrease saliency through stress and position

Who is mooing?The pig isnt.The horse isnt.The cow is. The cow is mooing.Uhoh! The horse is mooing. Silly horse.

18RecastsImmediate adult response to child utteranceRepeats some or all of the childs wordsMaintains the childs central meaning

Corrective: Corrects child error(s)

C: The cow mooing.A: The cow is mooing.

Non-corrective: Provides an alternate formC: The cow is mooing.A: Is the cow mooing?

ExpansionsImmediate adult response to child utteranceRepeats some or all of the childs wordsMay or may not correct a child errorMay or may not provide an alternate formAdds semantic content that changes the focus or shifts the childs meaningC: The cow is mooing.A: That mama cow is mooing for her baby.

MandsThe adult requests a response from the child QuestionsA: What is the cow doing? Requests for ImitationA: Say - The cow is mooing.

Direct InstructionExplicit instruction about the conditions for use of target forms.Adult: When there is only one animal say is. Here is one cow. When there are more animals, say are. Here are lots of cows.

What We KnowImitation > Models (Connell & Stone, 1992)Models > Imitation(Courtright & Courtright, 1976, 1979)Recasts > Imitation(Camarata & Nelson, 1992; Camarata et al., 1994; Nelson et al., 1996)Recasts = Models(Morgan et al., 1995; Farrar, 1990; Proctor-Williams et al., 2001)Recasts > Models(Farrar, 1992; Proctor-Williams et al., 2001, 2007; Saxton, 1997a; Saxton, 2000; Saxton et al., 1997) Models = Direct Instruction(Swisher et al. , 1995)

Conversational models & recasts work well for forms that frequently occur naturally Recasts must be used at least twice as often as in normal conversationTherapy models and imitation work well for forms that are rareImitation leads quickly to first use but not to generalization

23Inconsistent evidence because:Different types of studies, Different definitions,Techniques are not applied in their ideal way across studies, Different targets, Different initial levels of child accuracyFor example, in Connell & Stone, contrastive imitation vs receptive only models; in Courtright & Courtright, non-contrastive imitation vs. models with a production demand.In Proctor-Williams et al. we found no advantage of recasts over models for articles, but significant advantage of recasts over models for copulas

What We Dont KnowHow the most effective use of one technique compares to the most effective use of another techniqueWhether techniques are more effective when used in combination than in isolationIf combinations of techniques are more effective, which ones presented in which order?

Practice : am, is, are25Techniques to ProceduresWhen techniques are combined, they become procedures!

3 possible outcomes: The whole is greater than the sum of the partsEffects are additiveUse of one technique negates the effectiveness of another

the whole is greater than the sum of the parts. For example, the combination of setting up a communication temptation in a real context, providing a model, implementing time-delay to prompt an attempt, and imitation of the childrens correct productions or recasting their errors, could create conditions for what Nelson (1980) called a rare event. Rare events cause children to restructure some aspect of their immature grammar and adopt the correct form. As Nelson Denninger, Bonvillian, Kaplan, and Baker (1984) argued, the timing, quality, and context of the adult input may be as important as the presentation of a correct or an alternate form. Second, it may be that effects are additive. For example, a session could begin with a contrastive imitation/mand task, which would focus the childs attention and prime motoric skills for the target. The clinician could then introduce a more naturalistic activity that offers models and recasts to promote acquisition and generalization (Camarata & Nelson, 1992; Fey et al., 1993; Fey, Cleave & Long, 1997; Fey & Proctor-Williams, 2000). Finally, it may be that the use of one technique negates the effectiveness of another. For example, a child might find that a task that elicits frequent errors and thus allows high rates of corrective recasts followed with an imitation/mand and undue delay in awarding of a positive consequence is aversive in its similarity to nagging.

26Intervention Contexts & ProceduresChild-Centered Naturalistic, daily interactionsInvolves caregiversParticularly helpful for children who :Are responsive but with their own agendasRarely initiateStrong generalizationValuable during initial stages of language development

HybridClinician controls choice of activities and materialsChild controls play and conversational topicsStrong generalizationValuable for increasing exposure to rarely occurring formsUseful for early- and later-developing forms

Clinician-DirectedClinician controls all aspects of the treatmentRapidly elicits accurate task-specific productionsMust build in generalization tasksValuable for increasing exposure to rarely occurring formsUseful for early- and later-developing forms

Intervention Contexts & ProceduresChild-Centered Hybrid Clinician-DirectedHanen ProgramsPrelinguistic Milieu FastForWordTreatment

Milieu TreatmentEnhanced Milieu Drill/Drill-play Treatment

Dialogic ReadingFocused Stimulation Direct Instruction

Conversational Recast Treatment

Story Grammar

Procedures & Stages of Development

Rules of Thumb for Changing Goals and ProceduresTarget first words when children produce 2 communicative acts / minTarget early morphemes and simple sentences when children are combining 3 wordsTarget complex sentences when simple sentences are mostly grammatically correct20% of childrens sentences should be complex before kindergarten entryTarget stories when children use complete simple sentences and begin to string them togetherChildren should be telling personal event and retell stories that are mostly understandable before kindergarten entryRetell and stories will still often end-at-the-high point.

Dialogic ReadingTechniquesModelsMands: QuestionsRecastsExpansionsCROWD QuestionsCompletion /Cloze StatementsRecall QuestionsOpen-Ended Questions Wh-questions Distancing Prompts/ Questions

Dialogic reading was created with the belief that young childrens language development can be enhanced. It was derived from the theory that practice, feedback, and appropriately scaffold interactions produced positive effects in facilitating preschool childrens language development (Arnold et al., 1994). Adults are taught specific techniques, which provide models of language, while engaging preschoolers in the world of picture books. Completion An utterance spoken by the parent that facilitates the child to complete the question (e.g. The girl said that she wanted to sit in a _____?)Recall Questions that require the child to recall events from the book (e.g. Can you tell me something that the child wanted to do in this story?)Open-Ended - Statements that encourage the child to respond in his or her own words (e.g. What does this picture make you think of?)Wh-question Who, what, where, when, and why questions asked by the parent (e.g. Where did the boy find the dog?) Distancing Prompts Questions that encourage the child to analogize events from the story to real life events (e.g. Did you ever go fishing with your grandfather like this child did?)

31

Dialogic Reading ExampleC: He rolls to the right. He...R: Can you tell me something that Ollie does?O: What do you think he does inside his egg?W: What do Gossie & Gertie do to try and get Ollie out? D: I see ducks on the lake. Do you ever see any ducks?

32Drill-PlayTechniquesSlowingModelsMands: ImitationMands: QuestionsRecasts

HierarchyImitationModel-MandSpontaneousGeneralization

Drill-Play ExampleImitation PracticeI need sugar AND I need butter.I need flour AND I need eggs.I need eggs BUT I dont need rotten eggsI need peanut butter BUT I dont need chocolate chipsPlay the Game (Matching)

For Generalization: Make cookies

34Story GrammarTechniquesDirect InstructionMand: QuestionRecastExpansion

ApproachMultisensoryLiterature-basedMetalinguisticMetacognitive

Story Grammar ExampleCharactersSettingKick-off/ Initiating EventActions/AttemptConsequenceGREEN construction paper for Character, Setting, Kick-offYELLOW Construction for the attemptsRED construction paper for the direct consequence / resolution

What We KnowChildren with LI require more exposures to specific forms that they are ready to learn than are available in typical conversation to acquire morphosyntactic forms at the same rate as children with TLWhat We Dont KnowThe optimal frequency for using techniquesThe optimal techniques and their frequencies for different morphosyntactic structuresThe optimal techniques and their frequencies for children with different etiologies

What We KnowExpressive language outcomes are very similar for clinician- and parent-delivered intervention (Law, Garrett & Nye; 2004; Fey et al., 1993, 1997)This is as likely attributable to total frequency and distribution as it is to dose ratesWe can teach parents a wide variety of techniques and procedures (Girolametto et al., 1998; Hemmeter & Kaiser, 1994; Kaiser & Hancock, 2003; Kott & Law,1995; Wilcox 1992)What We Dont KnowWhat is the dose frequency and distribution that caregivers use in the home/classroom and can we measure this?How can we help caregivers sustain and adjust their dose frequency as the childs performance changes?The impact on children and their families when parents become intervention agents

38LAW ET AL. For children with expressive language difficulties only:No difference between clinician- and trained parent-delivered treatment (i.e., Hanen)as measured by no. of utterances, MLU, parent report of sentence complexity This is when children with only severe receptive language difficulties were excluded.For children with expressive language only, language intervention was better than no treatmentFEY ET AL. Parent-delivered intervention was as facilitative as clinician-delivered intervention, but only for those parent-child dyads who maintained higher rates of recasts without direct intervention designed to train parents of children with language impairments to use more recasts, their rates of total recast usage remain remarkably stable, and relatively low, over timeAfter 10 months of intervention, the parents whose children had made the most gains over the 5 previous months used approximately two recasts per minute in interactions with their children. In contrast, children who made small or no gains over this period received an average of less than one recast per minute. we began each individual session with contrastive imitation exercises. Ten models of one target form were presented first, followed by 10 models of the contrast target. After these models, the child was required to alternate targets, one after the other. We made no effort to continue this exercise to any learning criterion. Rather, we reasoned that this contrastive drill might focus the childs language learning resources on the target form. Following this 10-minute activity, we proceeded to play with the child in more naturalistic activities and provided opportunities to model the target form frequently and to recast the childs sentences using the target form.

What We KnowChildren more accurately produced and generalized a complex syntactic construction (e.g., It was the cup that the frog took) when exposed to it over 5 or 10 days than when exposed to it for 1 day (Ambridge, Theakston, Lieven & Tomasello, 2006) Children with TL (but not SLI) more accurately produced novel verbs when recasts were distributed across 5 sessions than when recasts were massed within 3 sessions (Proctor-Williams & Fey, 2007)

What We Dont KnowThe optimal distribution of dose frequency within and across sessions for:different morphological forms and syntactic frames for children with different etiologies Whether principles of distribution can be applied to techniques and procedures as well as specific targets39For Forms: in two experiments Ambridge, Theakston, Lieven, and Tomasello (2006) included 48 children, 3 years, 6 months to 5 years, 10 months and 72 children 4 to 5 years of age. They presented and elicited productions of 10 verbs in a complex past tense object-cleft construction (e.g., It was the cup that the frog took) and then tested its generalization to untrained verbs. They found that the children who received the exposures over 5 or 10 days rather than in 1 day reliably produced the trained and untrained constructions more accurately.For a technique + a form: recasts, irregular past tense - Proctor-Williams & Fey, 2007. What We KnowIntervention of more than 8 weeks seems more effective than those of less than 8 weeks (Law et al., 2004)Intervention of 4-12 weeks seems optimal (Nye, & Seaman, 1987)Intervention in the first 4.5 months resulted in greater gains than in the second 4.5 months (Fey et al., 1997)Children who attended a Head Start preschool more regularly produced more complex utterances and benefited more from LFC and LST (Justice, Mashburn, Pence & Wiggins, 2008)What We Dont KnowThe outcomes we can expect based on length of intervention The optimal length of treatment for different techniques and proceduresThe effects of classroom-based curricula and programs on child language outcomes - immediate and long-termThe consistency of attendance on individual treatment outcomes

40Measured by upper bound index (longest 5 utterances) and/or % complex utterancesBeing in attendance regularly had a stronger positive influence on childrens expressive language ability in Language Focused Curriculum classrooms compared with comparison classrooms and in classrooms exhibiting high rates of Language Stimulation Techniques compared with classrooms exhibiting moderate rates of LST use Proctor-Williams & Fey (2007)No. of days from the 1st to 5th experimental sessions ranged from 4 to 44 days; M = 14 days; SD = 8.95). No reliable correlations between length of time and accuracy of past tense productions in the low and high density conditions Based on effect sizes, for children with SLI, the longer they were in the experiment, the less accurately they produced the verbs. Spacing: Gaps of 5+ days between any visits did not affect the childrens verb accuracy

Effect sizes for clinician-delivered intervention increased when only studies of more than 8 weeks were includedThere is no evidence that our intervention for children who have severe receptive language problems.

Final Take-home MessagesEarly intervention worksWe have powerful techniques availableWe have many procedures that workWe need to continue to monitor children with early language difficulties to ensure catch-up language growth continues into acquisition of early literacy skillsWe need clinicians to work with researchers to answer what we dont knowProcedurePrelinguistic1st WordsSimple SentencesComplex SentencesStories

Preling Milieu Tx

Hanen

Mileu Tx

Focused Stimulation

Enhanced Milieu Tx

Fast ForWord

Dialogic Reading

Conversational Recast

Direct Instruction

Drill-play

Story Grammar