Language treatment in progressive aphasia & dementia

We are currently conducting a treatment study targeting the maintenance of language in Alzheimer's Disease (AD) and Frontotemporal Degeneration (FTD).  As these conditions worsen, people experience profound language impairments. These disorders have devastating consequences on a patient's functional independence and on caregiver well-being. We visit patients in their homes and craft a communication picture board specific to each person. We trained those words over time as language and cognition worsen.  

treatment approaches for language & communication in dementia

Most conventional language treatments work with the implicit goal of restoring lost function. This philosophy is often justified when someone suffers a stroke. However, the nature of a neurodegenerative condition makes restoration difficult.  Learning effects are often rigid and short-lived in the context of a memory disorder. An alternative aim is to pursue compensatory techniques such as structuring a patient's environment to optimize their cognitive and linguistic functioning. Memory books and video "banks" fall under this intervention approach.  We have proposed a technique centered upon maintenance of a set of critical words over a two year span. Our treatment works by cementing a set of 100 words during early stage dementia (within 1-2 years of onset) and continually training that same set of words over a span of two years. Thus, we work directly with the patient and caregiver to tailor a small, specialized vocabulary (a micro-lexicon). We then train those words using a technique we term, modified semantic feature analysis (mSFA).



How the treatment works

Our treatment works by training a core vocabulary of approximately 100 words.  Most people who enroll in the study already know most of these words, and that's great.  Our goal is to protect this small but essential vocabulary against the progression of Alzheimer's Disease and associated disorders.  A crucial component   of this treatment is how we select the words for training.

principles for selecting the words

Let's start with a thought exercise.  Imagine you have been banished to a desert island to be accompanied only by a robot. This robot will be your only communication partner for the next 50 years.  You have the option of programming your robot's limited memory with 100 words. Once you arrive at the island, however, the robot cannot be reprogrammed, and the robot cannot learn new words.  Which words will you select for your robot?

This seems like a very strange exercise.  However, we put a great deal of thought into it.  It helped us to generate principles for optimizing a small vocabulary.  We put these principles to work in generating lists of words to train in Alzheimer's Disease and progressive aphasia.  Here they are:  

  1. Target words must be highly familiar and highly frequent.
  2. Target words must be concrete (i.e., able to see, hear, touch).
  3. Target words must be highly functional and contextually appropriate.
  4. Target words must combine well and should be associated with each other.
  5. Target words must have high information content (i.e., no wasted words).
  6. Target words must represent an array of key semantic categories for daily living
    • People (loved ones), Hygiene, Clothes, Foods, Activities, Places

combining the training words into a cohesive system


We developed a communication board/book system using the above principles. The structure looks something like the picture on the right. We started with a big poster format but found that it was too distracting for patients to consider all 100 pictures at the same time. Now we use a book with four pictures per page

If you are curious what the items actually are, you can link to the lists in PDF form.  We generated item lists that are relatively gender specific for the category, clothes.  If you identify as male, here is your list  [download here] and female here is your  [download here].    

All of these words are rated and roughly matched across categories for familiarity and frequency using published norms.


training your target words

Our aim is to improve language. We also need to figure out who will be a good candidate for this type of language therapy.  Our first step is to get a handle on what the current level of language and brain functioning. In treatment research, these are called baseline measures. 


When you enroll in the study, we take a very clear picture of your brain using MRI.  It might look something like the brain on the left. At around the same time, we also do some standardized testing. Our tests examine naming, picture recognition, memory, and attention.  We also use a technique called eyetracking to examine how your eyes move when looking at pictures.   Now that we have a clear picture of your brain and now have a good handle on your language abilities.  Next comes the good part. We visit your home and take pictures of all the items on your specific word list. Then we edit your photos (crop to size, blur any text), and create a picture book for you depicting your 100 items (105 to be exact) .


Now it's time to start the treatment. We use modified versions of two therapy techniques called errorless learning and semantic feature analysis. We visit you at home and go through your word list with you and your significant other or caregiver. We use a matrix that looks like the picture to the right.

We train each word using a series of semantic features. You can think of semantic features as bits and pieces of information about the object.   We work through your word list in blocks of five items.  For each target word, we tell you the name and five things about it.  Then it's your turn. We give you the same block of five items and have you generate its name, its features. You will then use the word in a novel sentence. 

How to Enroll

If you are interested in learning more, please contact Dr. Jamie