Four Myths About Language Sample Analysis

 Why don’t you use language sample analysis in your practice? “It takes too much time.” “The results are difficult to interpret.” “It’s just hard to learn how to do it.”   “The whole process is too difficult.” Sound familiar? I promise you, it’s not as bad as you think. There are a lot of misconceptions and outdated ideas about language sample analysis floating around out there. So let’s take a closer look at these objections and, hopefully, debunk them.   Myth 1: LSA takes too much time After all, we all had to collect, transcribe, and analyze a language sample in grad school and it took hours and we hoped never to do it again. Well, yes, it takes forever the first time you do it, especially if you do it by hand. But it gets better. The truth is that there are new tools available that reduce the time LSA takes. That is the whole point of SALT Software in particular. Basically, SALT has developed time saving tools for every aspect of the process:

• Detailed elicitation protocols simplify recording a sample. You will know what to do and how to do it, making for a quick and smooth process. For example, a story retell exercise takes about four to five minutes on average using the SALT protocol.
• Transcription is simplified by using the SALT editor (and a few other tips - more on that later). The program helps you set up the transcript and provides help coding less frequent language features.
• Transcription conventions for coding all relevant aspects of oral language are available. Once you know the codes, putting them into the transcript is easy; and once the codes are in the transcript there are a plethora of analyses that SALT can run for you. If you understand the codes, you may not have to apply all of them to a transcript to obtain relevant analysis outcomes. More on that in a future post.
• Analysis of the transcript is fully automated with almost instantaneous results.
• Interpreting the results is aided by databases of typical speakers. If you elicit a language sample using the SALT protocols, it can be compared against age or grade-matched speakers, allowing you to easily spot (and empirically document) areas of strength or weakness.

I know most people dread it, but you really don’t need to be intimidated by transcription. We have a transcription lab and we can run the numbers for you. Our transcribers work part time; they do this regularly but they are not spending their whole lives doing this anymore than you would. We can tell you that, with practice, you can transcribe language samples at a rate of five to seven minutes per minute of recording. An average story retell takes four to five minutes and so can be transcribed in 25 – 35 minutes. How do you get there? A few suggestions (with shout-outs to a few of our favorite partners in transcription):

• Spend a little time up front learning the recommended coding protocols. The SALT website provides detailed online tutorials to learn the conventions resulting in 85-90% accuracy rate when completed.
• Get a digital audio recorder. Digital audio files are so very much easier to work with. And, truth be told, you probably already have a digital recorder - your smartphone.  Move the digital file onto your computer before transcribing so that you can use a foot pedal or other specialized audio playback options.
• Consider getting a foot pedal to control playback. You can use your foot to start or stop, speed up, slow down or back up, leaving your hands free to type. We like the Start-Stop Foot Pedal Transcription Systems; it is what our transcribers use.
• For a free option, try the Sound Scriber program for PC from Eric Breck at the University of Michigan to control playback of digital audio files. This program has a useful feature called "walking" which plays a small stretch of the file several times, then advances to a new piece, overlapping slightly with the previous one. No actual feet involved.

As you get more familiar with these tools, incorporating them into your standard practice, the time it takes to transcribe and analyze a language sample will be drastically reduced. All told, collecting and transcribing a sample could take as little as 40 – 50 minutes and the analysis is done in seconds using SALT Software. Yes, you will have to add the interpretation and the suggestions for intervention. But you would have to do that no matter what. And after all, that is why you entered the field to begin with: the challenge of figuring out individual communication deficits and what to do about improving everyday communication.   Myth 2:The results are difficult to interpret You are very familiar with the concepts of language impairment and a range of standardized tests to identify LI across age and grade. So let’s assume that you recorded and transcribed a language sample, then went into SALT Software to run some analyses and ran into information overload. There are tons of reports (most of them are under either the “Analysis” or “Database” menu) and each one spits out what looks like a mini-spreadsheet with way too many numbers. How are you supposed to interpret all this? Where should you start? Short answer: the SALT Performance Report. Let me first say that interpretation can never be completely automated; this is the core of your skill and value. But SALT Software has recently gotten around to helping out a bit more with the identification of significant measures and placing them in context. The latest version of our software includes the new Performance Report which will actually 1) flag significant results (any areas where your sample differs by one standard deviation or more - plus or minus - from the database averages), 2) generate a description of that measure for you, and 3) suggest further analyses that may be relevant for understanding each finding. From there you can do a deeper dive into just the most relevant areas. The SALT databases provide you with standard scores for each measure across word, morpheme, utterance and discourse measures comparing your client to age or grade-matched typical speakers. This should help in interpreting the results two ways. First, it brings the most significant results directly to the forefront for your consideration. Second, you can cut and paste the generated descriptions into reports to describe how each profile of measures captures a client’s spoken language performance. We hope that this will allow you to streamline analysis without losing either breadth or depth of information.   Myth 3: It’s just hard to learn how to do it Yes, it can be daunting. But the SALT team has developed materials to streamline the process of taking an audio recording and transforming it into an orthographic record; the program automates a ton of analyses, such as calculating measures of word, morpheme, utterance, discourse features along with noting errors, fluency and speaking rate. These tools allow you to document language use in everyday communication contexts. What could be better than that? So what we are really saying here is that transcription and coding are a little difficult to learn. Actually, more time consuming than difficult. And, yeah, that’s the worst part. But it gets easier with training and practice. First, take some of the free online courses to learn about transcription and coding. Then, the more you do it the easier it gets. Language sample analysis is one important tool (among many!) required to properly assess communication disorders, so it’s worth putting a bit of effort in. We are confident you can do this. The SALT program has gotten a lot more user-friendly and we are constantly making more training resources available to help. You have earned a master's degree; that was the hard part! So we are pretty sure you can learn this, too.   Myth 4: The whole LSA process is too difficult We take this argument to mean the whole process is overwhelming. And it may have appeared that way when you were a student or just beginning your practice, with limited experience with children and adults in assessment situations. But even if you have never elicited a language sample, you certainly have faced clients with communication deficits and have designed initial assessment protocols, administered tests following the administration guidelines, scored and interpreted the results following the test manual. You can do this. Take it in steps and you’ll see it’s really not that difficult.

1. Elicit a sample. Use this guide to choose the right protocol, grab your digital audio recorder or smartphone, and record a four to five minute sample. No problem.  
2. Transcribe the sample. Use this guide to get an overview of the transcription conventions SALT uses. Put the audio file on your computer; use a foot pedal or SoundScriber to control playback. Open the SALT editor and transcribe away!
3. Analyze the sample. Start with the Performance Report to identify significant measures for your specific client. You can find more resources here for help with analysis.