Tips for Eliciting a Valid Sample

Dressed in my “clinic outfit” in the modest room with the oversized two-way mirror, I’m ready to record. I have been instructed to elicit a conversational language sample from my client. My supervisor is behind that mirror. How hard can it be to get this child to talk with me? 

What I thought would be the easiest part of the session turned out to be a lot harder than I expected. What was supposed to be a conversation just wasn’t. I got plenty of constructive criticism (let’s call them “pointers”) after my session. Thank goodness I did. Once I learned what to do – as well as what not to do – I was able to get much better outcomes from my next clients, and plenty of data just ripe for analysis. Intentional, thoughtful language sample elicitation makes a difference. It’s one of the pillars of the LSA process: Elicit, Transcribe, Analyze, Interpret.

Now, years (and probably a thousand language samples) later, I can easily spot when a language sample goes rogue. Exposition digresses to conversation. Conversation morphs into narrative. It’s easier to go off course than one might think. But to get valid analysis results, we must stay the course. 

Examiners have specific roles in language sample elicitation. Those roles are designed to provoke the language features of importance for analysis. If discourse measures are of interest, a conversational sample will likely generate the best data. If you want to evaluate elements of story grammar, narrative samples will do a better job. In addition, if you plan to compare your language sample to a SALT reference database, it is important to closely follow SALT’s elicitation protocols. The further you stray, the less valid the comparison becomes. 

If we are focused on – and intentional with – how we collect a language sample, we will end up with a more valid analysis and save time. 

 

Joyelle Divall-Rayan, M.S., CCC-SLP
SLP: Vancouver, Washington
Director of Education & Training: SALT Software

 

________________

Here are SALT’s best guidelines for successful sample elicitation.

Establishing the groundwork.

  • Have a plan. Know what type of sample you want to elicit and follow the elicitation protocol specific to that sample type. Your role changes slightly between contexts, so take a minute to review the protocol and refresh your memory. In particular, know how much (or little) you should participate in the language sample (e.g., talking and turn-taking during a conversation or primarily active listening during a narrative).
  • It’s best to try and create an environment of trust. This may seem obvious, but taking a few minutes to establish rapport before diving into the task is time well-spent. 
  • Speakers are likely to say more if they believe the listener is interested in what they have to say, so you really gotta care about what they’re saying.
  • In some cases, the speaker may give some of the best they’ve got right out of the gate, before you have officially “begun” the sample. Maybe because it’s informal and there’s no sense of pressure. Maybe because there has been no mention that what is happening is a task, or worse, an assessment. Maybe there is some anxiety and talking is a coping mechanism. Whatever the reason, it’s a tricky situation. You don’t want to break the flow, but you need to share that you will be recording. Take a moment now to think about how you might handle this.
  • Use prompts only as needed. If the target speaker is going along just fine, don’t feel compelled to talk just for the sake of talking. If you sense more and better language can be elicited, you may need to prompt for it.
    • Non-verbal prompts (e.g., nodding and smiling) and/or non-specific verbal prompts (e.g., “uhhuh,” “keep going”) are almost always appropriate. More substantive prompts may or may not be allowable, depending on the context (more on that below). 
    • SALT’s protocol documents provide examples of appropriate prompts for each sampling context and are free for you to download and use. Remember that the prompts are there if you need them but are not part of the required script.

 

General Elicitation Tips: 

  • Be friendly and enthusiastic.
  • Be an active listener: give undivided attention, eye contact, vocal inflection, and positive body language.
  • Be patient and allow for “uncomfortable” silence. Silence can be difficult or awkward for typical speakers. However, some of the speakers we work with may need time to process, formulate, recall, and produce. Give them time for these facets of language.
  • Use a relaxed rate of speech. Fast can be construed as “let’s get this done,” and can cause communicative pressure. The target speaker’s processing speed should be taken into consideration as well. So, if you’re a “rapid-rater,” take it down a notch. 
  • Don’t interrupt or speak over the child. You’re missing out of what they might say! 
  • I offer that we listen to the sample after we’re done and kids love that! Unlike me, they seem to love to hear their own voice. 

 

If eliciting a personal narrative, story retell, expository, or persuasion:

  • When choosing between story retell and personal narrative contexts, ask yourself which you think will work best with this specific client: will she talk more if given a clear framework, or is she better off when she gets to choose a topic she’s interested in? 
  • Your role is to listen and show that you are interested in what the speaker is saying.
  • Ask for clarification if/when necessary and of value to your assessment. If the speaker can go it alone, that’s the better option. 
  • Use prompts of encouragement (e.g., “Uhhuh,” “keep going”) when needed.
  • Useful prompts include but are not limited to:
    • What else?
    • What happened?
    • Keep going.
    • Just do your best.
    • Mhm.
    • Uhhuh.
    • You’re doing great.
    • Tell me what you can.
    • That’s interesting. Tell me more [about that].

For personal narrative (not expository or persuasion):

    • Tell me more about that.
    • I don’t know about that. Can you explain / tell me about / describe ________?
    • “I wonder” statements.

 

When eliciting a conversation: 

  • Be mindful of how much you talk and what you are saying. Make sure you are taking your turn and leaving opportunities for your client to take hers.
  • If you are interested in responses to questions, then questions must be posed. If you want decent responses to questions, know what kind of questions to ask. 
    • Open-ended questions allow for longer, more thoughtful responses:
      • “Why…”
      • “How come…”
      • “How does…” 
      • “What about…” 
      • “What are some…” 
    • Yes/no and specific “WH” questions (e.g., “who” and “where”) can result in limited responses. 
  • Clients tend to speak more when they feel more comfortable with you. For some speakers, neutral questions that are not too personal are best. For others, getting a bit more personal (e.g., “tell me about [your dog] Rudy”) feels more comfortable and leads to more expression. If one style doesn’t work, try the other.
  • Try to refrain from asking questions that the speaker can infer you already know the answer to. It screams insincerity. Kids see right past it, too!
  • Allow enough time for your client to process language and answer the questions that were posed.
  • Prep the speaker by asking what they would like to talk about. I usually ask for two topics (e.g., pets or an upcoming event). 

 

If your sample does go rogue and you have a blended conversation/narrative/expository (who knows what) sample, you can always use the transcript-cut feature in SALT to restrict the utterances you want to include in analysis. Watch this video for more detail. Nonetheless, by being intentional about your sample elicitation you can save yourself a lot of time with less confusion, less transcription, and a more valid, robust language analysis! 

_______________

Check out our on-line courses for more information on how to elicit language samples.