| |
Abstract:
Japanese orthography utilizes Chinese ideographs, called
kanji, as well as two types of syllabaries, called kana. Kana
usually represent one mora each, while kanji can represent a
variety of different pronunciations which vary in the number of
morae. Consequently, Japanese phrases with the same number of
characters can differ in their length in morae.
In previous research, it has been claimed that the number of
characters rather than the number of morae is critical in reading,
comprehending, or lexical access of Japanese words in isolation.
From these results and others, it has often been claimed that in
reading Japanese, a lexical item can be directly accessed from the
characters of a word without necessarily accessing a phonological
representation of the word. However, few data are available as to
whether or not the same is true when a subject is reading a whole
sentence for comprehension. In this paper, relative contributions
of morae and characters are analyzed when Japanese sentences are
read silently.
In two eye-movement experiments and a self-paced reading
experiment, Japanese adults were asked to read various types of
sentences silently for comprehension. In the eye-movement
experiments, first pass fixation time and total fixation time (FFT,
TFT); first pass gaze time and total gaze time (FGT, TGT); and
overall reading time (TOT) were analyzed. In the self-paced reading
experiment, the reading time for each phrase (RT) was analyzed. For
each subject, each dependent measurement for each phrase (bunsetsu)
was submitted to linear regression analysis to determine the
coefficients for the relative contributions of morae and
characters.
The results showed that for most variables, both morae and
characters made significant contributions when they were entered
into the model alone. However, the number of morae and the number
of characters are highly correlated: in our experiments, their
correlation coefficients were about .73. Morae made significant
contributions even after characters were entered into the model
first. In contrast, characters were not significant when morae were
entered into the model first. A very similar pattern of results was
found in two eye-movement experiments in the majority of the
dependent variables. Mora was the dominant variable in the
self-paced reading time as well. These results suggest that in
reading Japanese sentences, it is the number of morae, rather than
the number of characters that largely determines how we read the
phrases. How many morae a phrase has, i.e., the phonological
representation of the phrase, seems to determine how long it takes
to read it even when the subjects were reading the sentence
silently. This is in contrast to the previous studies where words
were read in isolation. The results could be interpreted to suggest
that phonological representations of the words and phrases are
routinely accessed in normal reading, but there exists a direct
pass to access lexicon from the characters when a reader is
pressured to do so.
|
