Elton the dog, such a pretty boy!

My dog Elton, who puts the famous Marley to shame, is a frequent source of my embarrassment. His naughty acts are regularly followed by my apologies (and often payment) to whomever is the recipient of his deed. Because of Elton’s bad manners, I often find myself in socially awkward situations. Recently, after entering a neighbor’s garage and digging through the garbage for “treats”, Elton put me in yet another awkward situation. After I apologized for his deed, I wasn’t let off the hook. I didn’t receive the friendly, “Don’t worry about it.  No big deal. I love dogs.” What I did get was a blank stare, a slight frown, and a brief lecture about my ill-behaved dog. I left feeling dejected, embarrassed, and also thinking, maybe I should give obedience class one more try! I certainly was in the wrong, but the “social script” didn’t go how it usually goes (me apologizing profusely, Elton’s victim politely telling me, “It’s OK”). It made me think of those on our caseloads with social/pragmatic deficits or Autism Spectrum Disorders who chronically have a hard time “following the script” of expected social communication skills. How can I pinpoint pragmatic features and track progress on social goals in a valid, consistent, and meaningful way? I thought about how SALT would be very effective assessing the spoken language performance of people on the spectrum, or people receiving pragmatic services.

Elton the Dog standing on the kitchen tableWhen there truly is no harm-no foul, socially, the expected response to an apology is to accept it and move on. Both partners in the dyad are doing what is typical and called for. Let’s say we’re working with someone with poor social responses, i.e., pragmatic issues, who doesn’t produce expected responses. Taking language samples is an opportunity to assess and monitor change in the use of appropriate, inappropriate, or partially appropriate responses in any type of communicative interaction. The social situations from which samples are elicited can be contrived (role-played) or live, and performance can be functionally measured.

This is where custom coding using SALT is an invaluable utility. SALT’s standard Discourse summary does an excellent job providing data on common measures of conversation such as responses to questions, requests for clarification, number of turns, interruptions, overlaps, speaker turn length, and a variety of other valuable measures. But what if we want a finer assessment, one that is specific to the needs of our client? What if we get responses in communication interactions that are not socially acceptable, allowable, or required? For example, the generic nicety “How are you?” calls for a response such as, “I’m fine, thank you”, which is polite, be it truthful or not. Alternatively, we may get a response that delivers information indicating the speaker has more to say or wants it to be known things aren’t so good, e.g., “Things are going OK” or “Could be better”. These responses are socially acceptable. When implementing social skills intervention, using SALT, we can track performance for all types of social responses, acceptable/typical or not. With a few simple custom codes in transcription, all responses can be graded, so to speak. For example, you could set up a coding scheme that might include codes for appropriate response [AR], inappropriate response [IR], expected response [ER], unexpected response [UR], no response [NR], and maybe even a needs improvement response [NI]. Taking short conversational, situational samples over time and including custom codes in transcription will give you baseline data, data on change-over-time, and will direct you about where to focus your therapy? What types of questions, interactions, environments, and experiences elicit acceptable discourse or need more remediation and practice? Where, specifically, does the speaker break down? Where is the speaker most successful? Transcripts of short samples from real-life social interactions will provide valid data at the baseline and in future levels of performance.

The potential list of features to track and analyze is very large and very useful. A social communication coding-scheme might look like the following based on this goal:

“Jeffrey” will improve his social communication skills by:

  • Asking 3 appropriate questions of peers and adults within a two – three minute conversation.
  • Maintaining an appropriate topic with a peer or adult for a minimum of three conversational turns within a two-minute conversation.
  • Responding with appropriate (related) responses in 3 of 3 conversational turns in a two-three minute conversation.

By inserting a code on utterances from the target speaker, the resulting SALT analysis can provide baseline data, while subsequent sample analysis can track change over time. Intervention goals can be easily monitored and therapy can be focused accordingly. Below is a sample coding scheme and an excerpt from a conversation between a student and the SLP.

Custom Coding Scheme (developed for “Jeffery”):
[AQ] – appropriate question
[IQ] – inappropriate question
[TM] – topic maintained
[ATS] – appropriate topic shift
[ITS] – inappropriate topic shift
[TP] – topic perseveration
[AR] – appropriate response
[IR] – inappropriate response


Excerpt from a transcript of a conversation with “Jeffrey”

C What does Mrs_C teach?
C (Um) she teach/3s us math and reading [AR].
C Also we get lunch buddy/s [ATS].
E Who is your lunch buddy?
C We eat snack/s [IA].
C Whoever has the most point/s, we get to go with the teacher [ITS].
E Remember, I ask/ed who is your lunch buddy?
C Manny sometime/s.
E Can you tell me about something else you did this <summer>^
C <Oh yeah> [AR]!
C I play/ed video game/s [AR].
C (Uh) it stop/ed [ITS] {referring to bell}.
E That/’s OK.
E Just ignore it.
E So what else about summer?
C I went to my grandpa/z house [AR].
C And guess what?
E Hmm?
C Why is that there [ITS]?
E You can ignore that.
E Tell me about your grandpa/z house.
C He usually go/3s to work and work and work [AR].
C (He he he he) he only does the farmer thing/s [TM].
C But his wife (uh she has) she has a car [TM].
C And she take/3s him off the farm [TM].
E Oh, and where does he live?
C Well it was (:02 uh) poop [IR]>


SALT Analysis outcomes (PDF Version of Reports):

Snapshot of the SALT software Discourse Summary report Snapshot of the SALT software Code Summary report

Any feature that can reliably be captured can be custom-coded within a transcript of the communicative event. The benefit of going the extra mile is not only the data you were intent on getting, but also the multiple language features that SALT produces without the additional custom coding: number of total words, percent intelligibility, grammatical categories, and all the discourse features from the discourse summary. Those are just the tip of the iceberg. If you coded for bound morphemes, you would also get MLUw and MLUm as well as Number of Different Words. The list goes on. The gist is that custom coding a transcript of language can show present levels of performance and track progress on therapy targets. It can also show differences across contexts such as narration vs. conversation, or a mock job interview in person vs. a phone interview. There are endless possibilities for functional assessment and progress monitoring using custom coding schemes in SALT. Go ahead and try it! If Elton could talk, I’m sure the data would reveal he’s in need of some major pragmatic intervention!