For speech-language pathologists

How do you write AAC goals and use aided language modeling?

Published · Updated

The short answer

Aided language modeling means the communication partner points to symbols on the child’s AAC system while talking, so the child sees the system used the way they’re expected to use it. Systematic reviews and a meta-analysis, mostly of small single-case studies, found that interventions including it improved both comprehension and expression. Strong AAC goals cover more than vocabulary: Light’s four domains of communicative competence (linguistic, operational, social and strategic) give a useful frame, and training communication partners belongs in the plan.

What aided language modeling is

ASHA’s AAC Practice Portal groups several names under augmented input: natural aided language, aided language stimulation and aided language modeling. It describes augmented input as a receptive language approach in which the partner gives spoken words along with AAC symbols during communication, for example pointing to symbols while speaking. The idea is that language input provides a model for language development (ASHA).

In an early single-case study, five preschoolers who used AAC (three with voice output systems, two with communication boards) saw a partner point to two symbols on the child’s own system and then give a grammatically complete spoken model during play. Four of the five learned to produce multi-symbol messages consistently, kept using them and used them in new play routines; the fifth didn’t make consistent gains (Binger & Light, 2007).

For what AAC is, the kinds of systems and the worry that it stops speech, see what is AAC?. This page is about treatment.

What the evidence says

The caveats matter. Most studies are single-case designs with few participants, and modeling is usually one part of a package, so its separate effect is hard to isolate. The dosage figures come from small studies and aren’t a validated benchmark.

Core and fringe vocabulary

ASHA describes core vocabulary as high-frequency words that make up about 80% of the words most people use every day, and fringe vocabulary as lower-frequency words, mostly nouns, that tend to be specific to a context. Combining the two can increase how often AAC is used (ASHA).

The 80% figure comes partly from a study of 16 mothers talking with their typically developing infants between 9 and 15 months: just 267 words made up 80% of the 257,480 words they used, and the list overlapped substantially with the words toddlers and preschoolers say most (Quick et al., 2019). The authors suggest this helps decide which words to represent with symbols for aided input. One university program for students with significant disabilities builds its system around 36 core words, its Universal Core vocabulary (Project Core).

In practice, core words such as “go”, “more”, “not” and “help” can be modeled in almost any activity, while fringe words make the system personal, and families can help choose them. ASHA notes that lack of family involvement is a significant factor in device abandonment, which happens in about a third of cases (ASHA).

Writing AAC goals across the four competencies

Light proposed that communicative competence for people who use AAC depends on knowledge, judgment and skill in four interrelated areas: linguistic, operational, social and strategic. Linguistic and operational competence are about the tools of communication; social and strategic competence are about judgment in interaction (Light, 1989). ASHA’s portal adds psychosocial competence (ASHA). Goals that touch more than one domain stop a plan from becoming a list of nouns to touch.

Example AAC goals by domain (illustrations only; individualize criteria and contexts)
DomainWhat it covers (ASHA)Example goal
LinguisticKnowing and using the language, spoken and writtenDuring shared reading, the student will combine two symbols to comment (for example “go fast”, “not like”) in 5 opportunities per session for 3 consecutive sessions.
OperationalOperating the AAC system and techniquesThe student will navigate from the home page to the food page to request a snack, independently, in 4 of 5 opportunities across 3 sessions.
SocialKnowing what, when, with whom and how to communicateIn a 5-minute play routine with a peer, the student will use the device to take at least 3 turns, given aided models from the adult.
StrategicUsing available features to get messages across efficientlyWhen a listener doesn’t understand, the student will repair by choosing another symbol, gesturing or spelling, in 3 of 4 breakdowns.

Partners can have goals too, for example: “Classroom staff will provide aided models on the student’s system during snack in most opportunities.” For goal structure and wording, see writing speech therapy goals and the speech therapy goal bank.

Training communication partners

ASHA calls communication partners integral to AAC assessment and treatment, and notes that inadequate partner training has been identified as a barrier to device use (ASHA). The skills ASHA lists for partner training include active listening, longer wait time, augmented input, helping the user use rate-enhancement features, the operational side of high-tech devices, and supporting communication across settings (ASHA).

A meta-analysis of 17 single-case studies (53 participants) found communication partner interventions highly effective, with more evidence for children under 12, most with autism or intellectual or developmental disability. Aided AAC modeling, expectant delay and open-ended questions were the partner skills most often taught, usually through a descriptive overview, instructor modeling, guided practice and role play (Kent-Walsh et al., 2015).

Parents can be effective partners from the start. In a randomized trial of parent-coached interventions for 68 toddlers with fewer than 10 spoken words, the two augmented groups used both AAC and spoken words for target vocabulary and had larger vocabularies than a speech-only group (Romski et al., 2010).

The least dangerous assumption

Donnellan proposed the criterion of the least dangerous assumption in 1984, for the education of students with autism: when conclusive data are missing, base decisions on the assumptions that, if wrong, will do the least harm to the student’s chances of functioning independently as an adult (Donnellan, 1984).

Applied to AAC, the reasoning is that offering a robust language system and modeling it costs little if a child turns out to understand less than hoped, while withholding it costs a great deal if they understand more. It fits ASHA’s statement that there are no prerequisites for AAC intervention and that anyone whose communication doesn’t meet their needs is a candidate (ASHA).

It isn’t a reason to accept unvalidated methods. ASHA’s position is that facilitated communication is a discredited technique that should not be used (ASHA, 2018). Messages should be the child’s own, and progress should be measured.

Questions SLPs ask

What is aided language modeling in AAC?

The communication partner points to symbols on the child’s AAC system while saying the words aloud, during real interaction. It’s also called aided language stimulation or augmented input.

Does aided language modeling work?

Systematic reviews and a meta-analysis found interventions that include it improved comprehension and expression across ages and disabilities. Most studies are small single-case designs, and modeling is usually one part of a package.

How often should I model on an AAC device?

There’s no validated number. In small studies, modeling 70% of the time helped all children learn target words, while 40% worked for only some.

What are core and fringe words in AAC?

Core words are the small set of high-frequency words that make up most of what people say, like “go”, “more” and “not”. Fringe words are less frequent, mostly nouns, and specific to a person or activity.

What does “least dangerous assumption” mean in AAC?

It’s a principle from Donnellan (1984): without conclusive data, choose the assumption that does least harm if it’s wrong. In AAC that usually means offering and modeling a full language system rather than waiting for proof of readiness.

Sources

Join the Early Adopter Program