Retrieval-Based Word Learning in Autistic Children

There is a rich literature in cognitive psychology that highlights the importance of opportunities to recall and produce newly taught information when learning. The use of retrieval practice during a learning phase improves long-term retention between 50% and 150% relative to conditions without retrieval prompts. Recent work documents the power of retrieval practice in word learning studies with preschoolers with developmental language disorders. Though retrieval practice is not new, its application to child learning is in its infancy. Given the striking learning effects that have been documented in adults and emerging findings in children [with language impairment, retrieval practice has great potential to enhance learning in autistic children, a population with significant vocabulary deficits that often persist into adulthood.

Though autistic children have initial strengths in learning word form information (phonological), these word form advantages do not persist over time. Autistic children also demonstrate weaknesses in developing robust semantic representations, which have been noted by recalling fewer semantic features of newly taught words. Also, autistic children have difficulties in generalizing newly taught knowledge, which can hinder flexible vocabulary use. It has been suggested that autistic children may need more input, time, or practice than typically developing children to develop strong lexical representations.

In addition to deficits in breadth and depth of word knowledge, many autistic children demonstrate an atypical receptive-expressive vocabulary profile, with a reduced receptive advantage. Despite traditional therapy approaches recommending that receptive skills be targeted before expressive skills in autistic children, research indicates that it may be more effective to target expressive skills (i.e., word production) prior to targeting receptive skills. One recent study demonstrated that autistic children are more successful when learning words if they are prompted to name newly taught items rather than to identify (point to) a picture of these items. Targeting word expression led to more successful cross-modal generalization, with success extending to word comprehension (i.e., expressive-to-receptive generalization). This new finding holds promise for clinical practice; however, it has not been connected to a strong mechanistic explanation, which is necessary to strengthen the support for this practice. Retrieval-practice theory can provide the key mechanistic explanation that is required.

It has been proposed that effortful retrieval (now referred to as repeated spaced retrieval; RSR for clarity) enhances encoding because the retrieval of new information, or the attempt to do so, prompts individuals to identify features of the word or concept that are necessary to reconstruct the material, which leads to the development of an enriched and potentially elaborated memory trace to support future retrievals. RSR holds promise for autistic children because it may strengthen item-specific learning of phonological and semantic information (word form and meaning) that persists beyond immediate tests of learning. Furthermore, RSR may enable generalization (e.g., applying noun labels to new referents and adjectives to different objects). Thus, our specific aims are:

Aim 1: To determine whether repeated spaced retrieval (RSR) of labels (nouns) results in more robust learning of word form and meaning relative to a learning schedule that does not prompt retrieval.

In Study 1, 4- to 10-year-old autistic children will learn [novel nouns] in an RSR condition and a Repeated Study only (RS) condition. The investigators hypothesize that the children will recall more labels (word form) and semantic information (meaning) for RSR words and that their learning gains will persist at the 1-week test.

Aim 2: To determine whether repeated spaced retrieval (RSR) of adjectives results in more robust learning relative to a learning schedule that does not prompt retrieval. Study 2 will examine novel adjective learning, which has not been examined in autistic children. Autistic children will learn novel adjectives for unusual attributes displayed on familiar objects (e.g., "The cow is zogy."). It is predicted that the children will benefit from the RSR schedule when learning adjectives at immediate and 1-week tests.

Aim 3: To determine whether RSR enhances learning to the extent that children are able to generalize newly learned words to new images/referents. To demonstrate durable learning, the investigators will test beyond the explicitly taught stimuli. Study 1 will examine whether children will extend the labels (nouns) to referents that differ slightly in positioning and color. Study 2 will test for generalization of the newly taught adjectives when applied to objects from different categories (e.g., a zogy cow [taught] and zogy table [generalization]). It is predicted that generalization will be higher for words taught in the RSR condition than the RS condition.

Aim 4: To explore the moderating effect of autism severity on response to RSR. The investigators will examine whether autism symptom severity moderates word learning overall and if it predicts whether children benefit from RSR relative to RS learning schedules in Studies 1 and 2. The investigators will also explore associations between word learning and cognitive, vocabulary, and grammatical skills, and when warranted, statistically control for these variables.