Everyone speaks just a bit differently from others. Some of this is dialect-based; to see some great examples, take a look at the Harvard Dialect Survey, which has maps of such things as where people pronounce caramel with three syllables (“carra-mel”)vs. two (“car-ml”), where they say maple “sear-up” vs. “sih-rup” vs. “sir-rup”, etc. But some of these pronunciation variations are specific to the individual – as is the case in people with articulation difficulties.
Research in speech perception has shown that listeners “adapt” to the speech of the person they are listening to – they learn how that person talks and take that into account.
In a typical experiment, researchers create an ambiguous sound – something halfway between an “s” and an “sh”, say. We’ll call this sound “???”, since it’s ambiguous. This sound is then inserted in a variety of words – some words that should contain an “s” (such as absolute and research and castle) and some words that should contain an “sh” (such as potion and fashion). The researchers also create a continuum of sounds that range from the natural “s” through the “???” sound to the “sh” sound.
Study participants hear a list of either:
- the natural “s” words and these edited “sh” ones (research, fa???ion…), or
- a list of the natural “sh” words and these edited “s” ones (re???earch, fashion,…)
Then they hear syllables from across the continuum, and are asked to decide which items are “s” and “sh”.
What we find is that people change their perception based on that individual’s speech. So people who initially heard the ??? sound as being an intended “s” now think that more items in the continuum are an “s”. Those who initially heard the ??? sound as being an intended “sh” now think that more items in the continuum are an “sh”. That is, listeners change their perception in ways that adapt to the idiosyncracies of the talker – they stop hearing ambiguous items as ambiguous, and now hear them as being a normal “s” or “sh” from that speaker.
So why does this matter for SLPs? While everyone speaks slightly differently from one another, those individuals who speak “more differently” from normal (including those who produce ambiguous sounds that are hard to interpret) often get sent to an SLP – so therapists often work with individuals who speak in an atypical manner. And the goal in working with such individuals is clearly to make it easier for them to successfully communicate – to make their speech easier to understand. But how do you judge whether such therapy is working? Often, that judgement is based, in part, on our own personal listening to the individual.
But if we adapt to individuals’ speech idiosyncracies, such that they begin to sound normal for us, then our judgments are suspect — an individual is likely to sound easier to understand because we have adapted to them, even if their speech hasn’t improved from the perspective of anyone else!
That is, we might think therapy is working, when really it is our listening that changes rather than the client’s speech.
So this basic science research on speech adaptation leads to a very important insight that clinicians need to keep in mind – that interacting with someone in therapy can lead to changes in the therapist, not only to changes in the client. And in this case, those changes may make it more difficult to accurately evaluate improvement.
As therapists, there are things that can be done to help mitigate this effect when working with clients with speech sound or motor speech disorders or even those doing voice work. What can we do to help avoid overestimating a client’s progress?
1. Consider adaptation effects with each client
The first session with a new client gives us the opportunity to glimpse how intelligible or precise the client sounds to unfamiliar listeners. From that point forward, we are no longer an unfamiliar listener. Once we’ve spent some time with the client the adaptation effect will begin and we normalize the client’s pattern. Actively considering our tendency to adapt to speakers will help us take proactive steps to monitor for and minimize the effect of this phenomenon.
2. Take baseline recordings
Recording a speech sample from the first session will allow us to compare speech patterns before and after intervention more objectively. The grandfather passage and the rainbow passage are commonly used for this purpose. Consider also recording isolated words with target sounds since single words tend to have the lowest intelligibly. Functional phrases that the client says on a regular can be helpful for monitoring for changes in social or professional speech or voice. These recordings can also be a powerful tool that allows clients to hear their own progress.
3. Get a second opinion
Have a colleague or other professional listen to a speech sample. It is helpful to tell them as little about the client as possible in order to get more objective feedback. The input you want to elicit will depend on your goals. If you are working on intelligibility, you might ask for a transcription. For speech sound goals, you might ask for a list words or sounds that sound imprecise.
4. Compare other speakers
For certain clients, comparing speech samples to other typical speakers can help maintain objectivity when judging speech or voice naturalness or even listening for subtle differences in articulation. One of the easiest ways to do this is to record a speech sample using a standard passage, set of phrases or words. Listen to the client’s sample and samples of typical speakers to see if the client’s sample stands out.
5. Write intelligibility levels into your goals
When working with clients with decreased intelligibility, writing intelligibility criteria into goals is another way to remain vigilant and objective about ratings. For example, specify if a speaker must be intelligible to familiar listeners or unfamiliar listeners. It is also useful to differentiate between intelligibility, which is the ability to convey a speech message without context, and comprehensibility, which is when a context is provided. Clients, as with all speakers, are more comprehensible than they are intelligible. Other variables to consider? Predictability of the message, length of the utterance, and ability to see the speaker as the message is delivered. Getting specific about your criteria will help us all stay objective.
Kristin Slawson is a Clinical Assistant Professor at the University of Maryland. She is a licensed and certified Speech-Language Pathologist specializing in neurogenic communication disorders and voice. Her interests include overcoming barriers to individual social engagement, which includes vocal training for transgender clients and communication partner training for people with aphasia.