In 2014 Mirjam Blumenthal and Liesbeth van der Zijden began working on a concept for a novel speech assessment tool inspired by an existing informal test. With their colleagues at Royal Dutch Kentalis, they developed a screening tool to examine the speech development of multilingual children in their mother tongue.

What was your inspiration for Speakaboo?

While working at a local Audiological Center we noticed that for multilingual assessment it was frequently necessary to involve outside interpreters. Therapists don't always speak the patient's mother tongue. Analyzing spontaneous speech is time-consuming, costly and you can never be certain if the collected speech sample contains all the information you're looking for. We decided to make a device that would improve this process.

What was your starting point for creating this device?

We first looked at how frequently therapists required the help of interpreters, and what languages were commonly requested. Based on this information, we created a short list of languages that would be most useful for therapists in the Netherlands. In our initial selection, we excluded some languages, Mandarin Chinese for instance. Our lack of experience in working with tonal languages makes it too high a challenge for the starting phase of our project.

How did you select the target-words for Speakaboo?

For every language, we created a different set of words. Each compilation had to include all
consonants of the target language and had to be part of the vocabulary of young children. A list would include only nouns and consonants as target speech sound. We wanted to structure the test in a way that gave children confidence; we start off with easy to pronounce words and then gradually increase the difficulty. For each language, we did a literature search into the phonological system and acquisition order of consonants. If information about phonological development was not available (for Somali and Tarifit), we estimated the probable order of acquisition based on the frequency of the sounds and articulatory difficulty. We describe this process in more detail in our research paper.

Speakaboo tests are limited to an average of thirty words per list. Why this number?

When kids visit an Audiological Centre they engage in a whole range of tests. This can be tiring, particularly for young children with a short attention span. The thirty words translate to ten minutes of testing. Furthermore, the aim of the test is to get a first impression of the child's language and phonological development. Thirty words are enough to a achieve this objective.

How did you select the matching photos?

We would make the first selection based on our judgments. Next, we would evaluate the images with native speakers which often lead to changes. For instance, to represent the word 'woman' we once picked a photo of a young adult female. When we presented the picture to Somali individuals, the testers described the female as a girl. Consequently, to draw out the correct word during a test we had to replace the original picture with one of an older lady.

How do you handle dialectical differences?

We aim to create lists minimally affected by language variations. With words that either stay the same or demonstrate only minimal differences in dialects. If you look at the Tarifit word for ‘plate’ you can pronounce it as either 'topsi' or 'a-topsi'. Both versions test a child's ability to vocalize the combination of the consonants t, p, and s, so it has no negative influence on the testing procedure. We test all lists with native speakers. In cases with too many variations in how people name a photo, we skip an item. In some languages, the many variations between regions and sometimes even villages make it impossible to create a standardized set of words.

In your experience, how does the game component of Speakaboo affect the assessment?

We see that animations and interactive elements in a game help raise a child's level of engagement during a test. Often, Speakaboo is only one test in a line of assessments a child must undergo in one day. Kids quickly become bored. We introduce Speakaboo to them as a reward for completing the other tests. Young children enjoy the surprise effect of images popping out of a box. Older children familiar with games and tablets associate the devices with fun activities. They are eager to start using the application and enjoy the familiarity of swiping objects during the assessment. Therapists have also noticed that switching between the two tasks in the game helps improve the child focus on a test.

The underlying functionality of the program is another advantage for the therapist: it allows them to record per item, and later listen to and score a child's performance. This way they acquire a clear overview of the results. It also enables them to compare the speech of a young child with that of a pre-recorded interpreter.

An unexpected outcome is that parents who didn’t seem to understand what the assessment is all about, understand it better with the use of Speakaboo. A problem we have encountered in Audiological Centres is that some parents feel their children are examined for the wrong reasons because they are not fluent in a second language. Speakaboo makes it easier for guardians to understand that a child might also struggle it his or her mother tongue.

What comes next?

All our ideas will be put to the test during the coming months, starting with the upcoming beta. It will be our first occasion to test Speakaboo with a wider public. We are looking forward to seeing the application work in practice , and to the feedback from the users to help us improve the program. It would be great to collaborate in putting more languages in the application too. Also, the application may open possibilities for research.