LSA from conversation: an option for preschool and elementary

We’ve chatted before about how many SLPs are not using language sample analysis (LSA) as often as they should, and provided some examples of LSA tools to use. One of the LSA options previously discussed was the Systematic Analysis of Language Transcripts (SALT), which many of you used in graduate school, and some continue to use. SALT is very much a standard in our field; “…arguably the largest child language database available…” and “… a valuable clinical tool…” However, “…fewer than 25% of SLPs reportedly use SALT (Pavelko et al., 2016),” with barriers including cost of the software, time to do transcription and analysis, and “limited recognition as a valid assessment measure,” despite evidence to the contrary.

Then, there’s the SUGAR protocol, which has been designed with clinicians’ reported barriers in mind. So, what makes it different than other LSA protocols? SUGAR is:

  • Brief. Based on the current paper’s sample of 385 typically-developing children, ages 3–8, the entire LSA process (collect the sample, transcribe, analyze) took an average of only 20 minutes (range 10–33 minutes).
  • It’s based on 50 utterances only, with just four analyzes:
    • MLU (mean length of utterance): make sure to follow the authors’ directions on how to calculate MLU, because there are several ways to calculate MLU. The authors state, “….we wanted a way to calculate MLU that would be sensitive to age-related changes in both preschool- and early school-age children… because researchers have demonstrated that MLU growth slows considerably after age 4 years (Rice et al., 2010)…”
    • CPS (clauses per sentence): not clause type, but simply number of clauses. So you’ll have to identify the clauses out after transcribing.
    • WPS (words per sentence)
    • TNW (total number of words)
    • The authors chose these four analyses because previous research indicated that they would be sensitive to developmental change. Their study results confirm this—“…significant age-related changes in each of the four LSA measures…”
  • Elicited from conversation, which is reported by SLPs to be the most common method of eliciting a language sample. Previous research has shown than conversational language samples don’t always elicit as complex of language as expository tasks. However, the authors address this concern by providing specific instructions meant to guide the conversation in a way that elicits more narrative-style language.

Note that this study is entirely on typically-developing children. You may use this data to make comparisons, per analysis and per age group, for clients on your caseload. However, do note that the amount of data here and the options for analysis aren’t comparable to what the SALT database offers. However, that’s simply the cost/time tradeoff we’re working with here. The point of the SUGAR protocol is to give SLPs something that is compatible with clinical reality, that directly addresses the barriers that have prevented widespread detailed LSA use in the past.

So, how do you implement SUGAR elicitation, transcription, and analysis?

The authors have laid out everything in the appendices. The transcription and analysis don’t require learning a coding scheme, but instead take advantage of counting functions already built into Microsoft Word. I don’t know that it can get much easier than this (until somebody smart builds us a really good LSA app…). And, for the SLPs who think—“Nah. I’ll just count whatever makes sense and elicit the language sample however it plays out naturally…”, please note that this makes it impossible to compare data from your language sample with available research. If you don’t elicit and analyze the way the researchers did, you can’t make a direct comparison with their data set. Also, cherry picking of analyses that seem like they could be appropriate or that you happen to be comfortable with is also discouraged. It’s common for some SLPs attempt to make the LSA process shorter by analyzing just one metric (such as MLU), but the authors state that this is, “… not encouraged.”

Pavelko, S.L., & Owens, R.E. (2017). Sampling Utterances and Grammatical Analysis Revised (SUGAR): New Normative Values for Language Sample Analysis Measures. Language, Speech, and Hearing Services in Schools, 48, 197–215.