THESIS
2017
xv, 187 pages : illustrations (some color) ; 30 cm
Abstract
Past studies in English and German texts have shown that a vertical striped word (one
which consists of repetitive vertical stripe components) is associated with a longer
reading and binocular fixation times. A biologically inspired reason related to binocular
fixation has been proposed, which predicts that the prolonged reading time is caused by
the additional effort from the binocular vision system to align an image with a high
occurrence of vertical stripes projected on the left and right retinas. This study aims to
investigate if the proposed theory is true in reading Chinese. In the first experiment, we
planned to study whether similar effect persists in Chinese texts. Unlike English and
German texts, Chinese text are not alphabetical but are logographic. Three thousand
co...[
Read more ]
Past studies in English and German texts have shown that a vertical striped word (one
which consists of repetitive vertical stripe components) is associated with a longer
reading and binocular fixation times. A biologically inspired reason related to binocular
fixation has been proposed, which predicts that the prolonged reading time is caused by
the additional effort from the binocular vision system to align an image with a high
occurrence of vertical stripes projected on the left and right retinas. This study aims to
investigate if the proposed theory is true in reading Chinese. In the first experiment, we
planned to study whether similar effect persists in Chinese texts. Unlike English and
German texts, Chinese text are not alphabetical but are logographic. Three thousand
common Chinese characters (taken from a Hong Kong Baptist University Language
Center report) were analyzed into three groups: (i) character with more repetitive
vertical stripes; (ii) character with less repetitive vertical stripes; and (iii) neutral, using
horizontal autocorrelation as an objective measurement. Three measures were utilized
in the categorization process based on horizontal autocorrelation; (i) the value of first
peak of the horizontal autocorrelation; (ii) difference in value of first peak and the value
of the preceding trough in horizontal autocorrelation; (iii) number of peaks of horizontal
autocorrelation. The results from experiment 1 and 2 have shown that a character with
more repetitive vertical stripes exhibits a significantly longer recognition time than a
character with less repetitive vertical stripes. This suggests that the effect of vertical
stripes is not restricted on alphabetical texts. In the third experiment gaze movement
were recorded throughout to identify the binocular fixation duration for each character.
Results suggests that the more repetitive the vertical stripe pattern was, the longer the
binocular fixation durations. In summary, the presence of vertical stripes prolonged
recognition time and binocular fixation duration of Chinese characters. According to the
proposed theory, in a monocular viewing condition the recognition time and fixation
duration shall not be affected by the presence of vertical stripes. In addition, the
presence of vertical stripes should affect the recognition of a character regardless of the
reading orientation. To put the proposed theory under the test, the fourth experiment
studied 3 viewing conditions; (i) horizontal binocular viewing; (ii) vertical binocular
viewing; (iii) horizontal monocular viewing (non-dominant eye to be occluded). As
predicted by the theory, recognition time for character with more repetitive vertical
stripes is significantly longer in condition (i) and (ii), but not in condition (iii). Binocular
fixation duration (while in condition (iii) it would be monocular fixation duration), is
significantly longer with character with more repetitive vertical stripes in condition (i) and
(ii) but not in condition (iii). The fifth experiment utilized an iCub simulator and a
biologically inspired binocular vergence model to simulate fixation on 288 Chinese
characters with different levels of repeated vertical stripes. Results indicated that the
model does reached a smaller binocular disparity viewing a character with more
repetitive vertical stripes. This supports the proposed theory.
Post a Comment