Ever tried to take a language sample with a child who uses AAC? You probably experienced a few challenges along the way. Read on for how to address two major challenges of language sample analysis with children who use SGDs (speech generating devices).
Challenge #1: Obtaining a valid and representative language sample. Twins who use SGDs participated in this study. Researchers had access to existing language data collected automatically by the participants’ SGDs across multiple years, at 7;3 years old and also age 8;5–12;5.
The researchers found that a one-day sample window did not provide enough utterances for analysis. A one-month sample window consistently provided a sample of more than 50 multimorpheme utterances needed for analysis. For children who use AAC, if the target is 50 multimorpheme utterances, you may need to adjust the length of your current sample window in order to obtain a representative sample using data tracked automatically by the SGD.
Challenge #2: Transcribing short utterances out of context. If you have a string of single-morpheme noun utterances, and no context for the conversation, it can be difficult to interpret where one utterance begins and another ends. Enter: Mean Syntactic Length (MSL). MSL is the average number of morphemes per utterance, excluding one-morpheme utterances. Kovacs and Hill propose using MLUm (mean length of utterance in morphemes) when the context is known; however, MSL appears to have promise for decontextualized samples. Take a look at Figure 2 in the study for a list of rules to assist you when calculating MSL.
Note: This is a small study, not meant to prescribe what to do broadly, for children who use AAC. Rather, it provides clinicians with some evidence-based options to consider within their clinical practice.
Kovacs, T., & Hill, K. (2017). Language samples from children who use speech-generating devices: Making sense of small samples and utterance length. American Journal of Speech-Language Pathology, 26, 939–950.