This post is also available in Dutch.
Nowadays we are flooded with new technologies and inventions. Some of these have become our heroes in daily life (e.g., mobile phones, PCs, the internet), while others are less useful and/or just need more time to improve (e.g., hairbrushes that send information about brushing quality to its users’ mobile phone via Bluetooth). A new invention that is gaining more and more popularity is so-called sign gloves – wearable technology that promise to translate sign language to speech in real time (for example these). Let’s have a look whether this product can live up to the promises made.
How does it work and is that good enough?
The gloves consist of five flex sensors mounted on each finger to detect if and how much a finger is bent. The glove is then connected via Bluetooth to, for example, a phone where the signs are translated into speech or text. That’s the theory. In practice, these are rather clunky proto-types that misconstrue the nature of sign languages by only detecting what the fingers do (and even then not in great detail). Key features of signs are not only how much a finger is bent but consists of other features such as hand movement (e.g., up, down, circular, straight), palm orientation (e.g., palm points toward the signers body or away from it), and the location (e.g., in front of the chest, at the side of the head). Besides, even if the gloves could track those features perfectly (currently they don’t), they ignore other key parts of sign languages such as facial expressions (which can have grammatical functions), the signer’s torso, or mouth movements. Moreover, signs are not simply produced sequentially like pearls strung on a thread (pearl after pearl) but rather in a more simultaneous way combining many key parts together at the same time. All in all, this complexity of sign languages creates a huge challenge for machine sign-to-speech translation (or vice versa).
Should we even strive for a sign language glove?
Sign language gloves were invented to facilitate communication between the Deaf and Hearing. However, these gloves only translate sign to speech which is a one way street that aids the Hearing to understand the Deaf, while the Deaf still struggle to access speech. Moreover, it is the Deaf that need to buy such a device and also wear these rather clunky gloves to facilitate this one way communication. That doesn’t sound really appealing at all, does it? Thus sign language gloves are rooted in the disinterest of the Hearing, not in the actual needs of the Deaf. This is a consequence of not including Deaf people in the development and asking them what they really need. Generally, as long as the actual users (here: the Deaf) aren’t included in the development, new usage-based technologies will fail to actually aid the target user.