Display of Virtual Braille Dots by Lateral Skin Deformation: Feasibility Study
table of contents5. Control Experiment
The reading task performed in the legibility experiment was not representative of typical Braille reading. It was hypothesized that the reading difficulties experienced by some subjects were inherent to reading a single row of Braille dots. The dots found in the bottom rows of most Braille characters could facilitate the localization of the dots within the cell. Without this extra information, locating a dot completely depends on evaluating the length of the spaces between dots.
A follow-up experiment was designed to test the subjects' ability to read single-row Braille characters on a conventional Braille medium: Braille embossed on vinyl tape.
5.1 Method
5.1.1 Participants
The experiment was conducted with the same five participants, one year after the original experiment.
5.1.2 Materials and Tasks
The reading task was identical to that used in the original experiment. Subjects were asked to read sequences of four single-row Braille characters starting and ending with '
'. The Braille strings were embossed on 1/2" adhesive vinyl tape using a Braille labeler. The non-existent 'dot #4' character was produced by sanding down the extra dot on character '.' of the embossing wheel.
The resulting Braille has smaller, sharper dots than paper Braille but is still easily readable and commonly used. Vinyl was preferred over paper because it afforded better control on the uniformity of the test plates.
Sixteen Braille labels (one per string) were produced. Each was affixed to a thin, right-angled metallic plate. The placement of the tape was carefully controlled to minimize differences between the plates and allow sufficient space for the finger. During trials, the plates were held down by a switchable magnetic clamp that allowed us to change the strings quickly, see Figure 16(a). A single flattened dot was printed on the extreme-left of the tape to serve as a starting point for reading.
5.1.3 Procedure
The subjects were read written instructions and had supervised practice trials until they felt comfortable with the task. They were presented with strings to read in block trials. They placed their finger to the left of the plate, waited for an audible signal, slid their finger over the flattened positioning dot, read the string, reported verbally the two middle characters, slid their finger back over the tape, and removed it from the plate. They were instructed to read only with their right-hand index finger and to keep their other fingers against their palm. There was no time limit but they were strongly encouraged to answer quickly. In case of doubt, they were asked to give their best guess. The experimenter logged the result of each trial. A trial block comprised 80 strings with each of the 16 possible combinations appearing 5 times in randomized order.
5.1.4 Data Collection
A camera was positioned above the reading surface and was used to record color movies of the finger movements at a rate of 30 frames per second. The movies were compressed and stored for later analysis. A single, uncompressed image of the plate was also taken. The reading patterns and trial durations were analyzed from the movies captured during the experiment. Simple image processing operations were applied on the plate image to extract the position of the leftmost dot of the 4-character string. The absolute difference between the saturation levels of each frame with the background frame was then used to locate the fingertip in the image sequence. In order to approximate the definition of trial duration used in the original experiment, the trial was considered to begin as the rightmost part of the index crosses over the leftmost dot, and to end when it crosses it again in the reverse direction for the last time. The automated measurements were inspected visually and corrected for 75 of the 400 trials. The precision was estimated to be within 3 frames (±0.1 s).
5.2 Results and Discussion
5.2.1 Legibility
All five subjects read the 80 strings presented to them with 100% accuracy. It is thus clear that the reading difficulty experienced by the subjects with the VBD cannot be explained solely by the inherent difficulty of the task.
5.2.2 Reading Patterns
Table I shows the timing measurements. The subjects read faster on vinyl Braille than on the VBD, more than twice as fast on average. However, one of the slowest subjects on the VBD (AB) is also one of the fastest on vinyl Braille. The reading patterns were also inspected visually to assess their naturalness. Some subjects clearly slowed-down when sliding over dots and frequently returned to previous dots, or to the beginning of the string. This suggests that the reading strategies used by the subjects were different from those used in normal Braille reading [Millar 1997, Schiff and Foulke 1982, Bertelson et al. 1985].
5.2.3 Verbal Reports
Most of the subjects mentioned spontaneously that the reading task was easier on embossed plastic than it was on the VBD. Upon questioning, however, some acknowledged that the reading task was more difficult than typical Braille reading due to the lack of context (meaningless strings) and the absence of cues on the bottom rows. Subjects were also uncomfortable reading with a single finger, and particularly with keeping the other fingers in a fist.