詞視覺複雜度分佈對閱讀中文句子的眼跳標靶之影響 - 政大學術集成

全文

(1)國立政治大學心理學研究所 National Chengchi University Graduate Institute of Psychology. 碩士論文. 政治大. Master’s Thesis. 立. ‧ 國. 學 ‧. 詞視覺複雜度分佈對閱讀中文句子的眼跳標靶之影響 The Effect of Word’s Visual Complexity Distribution on Saccade. y. Nat. n. al. er. io. sit. Targeting in Reading Chinese Sentences. Ch. i n U. v. e n :g孟威廉研究生 chi. 指導教授 : 蔡介立. 中華民國一百零四年七月 July, 2015.

(2) 詞視覺複雜度分佈對閱讀中文句子的眼跳標靶之影響 The Effect of Word’s Visual Complexity Distribution on Saccade Targeting in Reading Chinese Sentences. 研究生 : 孟威廉指導教授 : 蔡介立. Advisor : Jie-Li Tsai 政治大. 國立政治大學. 學. ‧ 國. 立. Student : William Cruz Molina. 心理學研究所. ‧. 碩士論文. er. io. sit. y. Nat A a Thesis submitted to the. n. iv l C n Graduate Institute h e n g cofhPsychology i U National Chengchi University. in partial fulfillment of the requirements for the degree of Master in Psychology. 中華民國一百零四年七月 July, 2015.

(3) Table of contents. List of tables..................................................................................................................... IV List of figures .................................................................................................................... V Abstract ............................................................................................................................ VI Chinese abstract ........................................................................................................... VIII Acknowledgement ........................................................................................................... IX Forewords .......................................................................................................................... 1 Chapter 1. Literature review ........................................................................................... 5. 治政大Targeting ............................... 6 1.1.1 The Influence of Low-level Features on Saccade 立 Word Length ................................................................................................................ 6. 1.1 Eye Movement Studies in Reading Languages other than Chinese ..................... 6. ‧ 國. 學. Launch Site ................................................................................................................ 10 Lack of Interword Spaces .......................................................................................... 12. ‧. 1.1.2 The Influence of High-level Features on Saccade Targeting............................. 16 Morphological Traits ................................................................................................. 16. Nat. sit. y. Word Predictability .................................................................................................... 19. io. The Chinese script.............................................................................................. 20. al. n. 1.2.1. er. 1.2 Eye Movement Studies in Reading Chinese ......................................................... 20. i n U. v. 1.2.2 The Influence of Low-level Features on Saccade Targeting ............................. 22. Ch. engchi. Word Length .............................................................................................................. 22 Insertion of Interword Spaces .................................................................................... 27 Launch Site ................................................................................................................ 32 Font size ..................................................................................................................... 33 Visual Complexity ..................................................................................................... 34 1.2.3 The Influence of High-level Features on Saccade Targeting............................. 39 Word Frequency......................................................................................................... 39 Word Predictability .................................................................................................... 41 1.2.4 Saccade Targeting Strategies in Reading Chinese ............................................. 42 Word-based Strategy.................................................................................................. 42. i.

(4) Character-based Strategy ........................................................................................... 44 Fixed-amplitude Strategy ........................................................................................... 46 Flexible-saccade Strategy .......................................................................................... 47 1.2.5 The Visual Complexity Distribution Index........................................................ 51 Chapter 2. The effect of words’ VCD on saccade targeting ........................................ 61 2.1 Research Question .................................................................................................. 61 2.1.1. Hypotheses ......................................................................................................... 61. 2.2 Method ..................................................................................................................... 66 2.2.1. Participants ......................................................................................................... 66. 2.2.2 Apparatus ........................................................................................................... 66. 政治大 Design and Procedure ........................................................................................ 71 立. 2.2.3 Materials ............................................................................................................ 67 2.2.4. 2.2.5 Data analysis ...................................................................................................... 74. ‧ 國. 學. 2.3 Results ...................................................................................................................... 77 2.3.1. Duration measures ............................................................................................. 81. ‧. 2.3.2 Position measures............................................................................................... 81 2.3.3 Probability measures .......................................................................................... 82. y. Nat. sit. 2.3.4 Linear Mixed Models ......................................................................................... 83. al. er. io. 2.4 Discussion................................................................................................................. 84. v. n. Chapter 3. The VCD dependency on word representation ......................................... 91. Ch. i n U. 3.1 The Effect of Non-words’ VCD index on Saccade Targeting Mechanism ........ 91 3.1.1. engchi. Data analysis ...................................................................................................... 92. 3.1.2 Results ................................................................................................................ 94 3.1.3 Discussion .......................................................................................................... 96 3.2 The Effect of VCD index on Saccade Targeting Mechanism .............................. 98 3.2.1. Data analysis ...................................................................................................... 98. 3.2.2 Results .............................................................................................................. 100 3.2.3 Discussion ........................................................................................................ 102 Chapter 4. General discussion ..................................................................................... 106 4.1 The VCD Effect on Saccade Targeting Mechanism .......................................... 106 4.2 The VCD Effect on Word Identification ............................................................. 111 ii.

(5) 4.3 Limitations and Future Research ........................................................................ 112 Appendixes..................................................................................................................... 118 A. Text features of different language scripts .............................................................. 118 B. Mean of Chinese words’ VCD index for groups defined by total stroke count ...... 119 C. Descriptive Statistics of the reading material by VCD condition and syntactic category ........................................................................................................................ 121 D. Complete models on eye measures for target words............................................... 123 E. Complete models on position and probability eye measures for non-words........... 124 F. Complete models on position and probability eye measures for non-targets .......... 125 Glossary ......................................................................................................................... 127. 政治大. References ...................................................................................................................... 131. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. iii. i n U. v.

(6) List of tables. Table 1. Mean ILP by word length and language ............................................................... 9 Table 2. Mean ILP in different studies testing the morphological structure of words ...... 18 Table 3. Mean ILP in previous studies for different Initial fixation cases. ....................... 32 Table 4. Mean of Chinese words’ VCD index for groups defined by total stroke count and examples of VCD bias ........................................................................................................ 56 Table 5. Descriptive statistics of the computed VCD for different languages ................... 57 Table 6. Mean of word traits for Target, Pre and Postarget words .................................. 68 Table 7. Mean of word traits for Target, Pre and Postarget words .................................. 69. 政治大 Table 9. Estimated effects on eye movement measures...................................................... 80 立 Table 8. Descriptive statistics of eye movement measures by condition ........................... 78. Table 10. Descriptive statistics of visual features for target and non-word VCD conditions. ‧ 國. 學. ............................................................................................................................................ 93 Table 11. Estimated effects on probability and position measures for non-words ............ 96. ‧. Table 12. Descriptive statistics of visual features for target and non-target VCD conditions ......................................................................................................................... 100. y. Nat. sit. Table 13. Estimated effects on probability and position measures for non-targets ........ 102. al. er. io. Table 14. Descriptive statistics of eye movement measures for target, non-word and non-. n. target by VCD conditions................................................................................................. 104. Ch. i n U. v. Table 15. Mean ILP in 2-character Chinese words for different Initial Fixation cases . 108. engchi. Table 16. Mean words’ VCD index for target words and 2-character words in different font types .......................................................................................................................... 116. iv.

(7) List of figures. Figure 1. Cross language validation of the PVL phenomenon ............................................ 8 Figure 2. Mean ILP by word length distinguishing incoming saccades from different launch sites, data from McConkie et al. (1988) ................................................................. 10 Figure 3. Spacing conditions in Rayner et al. (1998) and Sainio et al. (2007) ................. 14 Figure 4. PVL curves from previous studies in reading Chinese. ...................................... 25 Figure 5. Spacing conditions used in Bai et al. (2008) and Shen et al. (2012) ................. 29 Figure 6. Three different font types for displaying Chinese characters ............................ 35 Figure 7. Example of character and 2-character word zones in the horizontal, vertical. 政治大 Figure 8. Visualization of the words’ VCD Index computation ......................................... 52 立 and quartile analyses from Zang et al. (2013) ................................................................... 37 Figure 9. Density distribution of words’ VCD index for different language cases ........... 59. ‧ 國. 學. Figure 10. Boxplot of words’ VCD index for different language cases ............................. 60 Figure 11. Experimental hypotheses on the distribution of initial fixations by VCD. ‧. conditions ........................................................................................................................... 64 Figure 12. Example sentences of the VCD conditions ....................................................... 71. y. Nat. sit. Figure 13. Diagram of the experiment procedure. ............................................................ 72. al. er. io. Figure 14. Mean of eye movement measures by VCD condition ....................................... 79. n. Figure 15. Boxplot of the VCD index for words with different character composition. .... 87. Ch. i n U. v. Figure 16. Visualization of the procedure for selecting Non-words .................................. 93. engchi. Figure 17. Visualization of the procedure for selecting Non-targets................................. 99 Figure 18. Mean of eye movement measures for target, non-word and non-target by VCD conditions ......................................................................................................................... 105 Figure 19. Description of an experimental design for testing the VCD index’s influence on saccade targeting under semantic inconsistency ............................................................. 115 Figure 20. VCD index computation for words displayed vertically................................. 117. v.

(8) Abstract. Previous studies about the visual and linguistic factors that influence the decision about where to move the eyes next in reading suggest a strong influence from low-level features; some studies also assume that this decision involves the computation of a saccade target before the oculomotor program is executed. In order to test whether the distribution. 治政 devised a new parameter that reflects the distribution大 of visual information along 2立. of visual components within Chinese words influence the saccade targeting mechanism, we. character words’ area: the Visual Complexity Distribution (VCD) index. Three groups of. ‧ 國. 學. words with a marked VCD index (i.e. Left-Bias, Right-Bias and Non-Bias) were identified. ‧. and embedded in natural sentences; the eye movement of Chinese native speakers was recorded while they read this material in order to contrast first-pass duration, landing. y. Nat. er. io. sit. position and probability eye movement measures between conditions. The experimental effects were estimated through contrast between conditions using Linear Mixed Models,. n. al. Ch. i n U. v. thus providing evidence about the VCD index’s influence on both, the decision about the. engchi. when and where to move the eyes next. The analyses on initial fixation position indicate a rightwards shift when sending the eyes towards words with Right-Bias in comparison to the other conditions and shorter fixation durations when biased words are fixated in comparison to the Non-Bias words. Further analyses demonstrated that the results above can only be observed when specifying saccade targets from Chinese words. These results indicate that the luminance patterns within 2-character Chinese words, as reflected by the. vi.

(9) VCD index, can influence the specification of a saccade target when those words are about to be fixated as well as modulate the fovea load when those words are currently fixated.. Keywords: Saccade Targeting Mechanism, Landing Position, Reading Chinese.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. vii. i n U. v.

(10) Chinese Abstract. 摘要。探討閱讀時視覺與語言因素之研究指出眼球移動位置的決定主要受到低階視覺特徵的影響。有些研究也認為此一涉及計算眼跳目標的決策發生在執行眼球移動之前。為了檢視中文詞彙內的視覺複雜度分佈是否影響眼跳目標決定機制、我們提出了視覺複雜度分佈指標 (visual complexity distribution index, VCD index) 來. 政治大. 代表中文雙字詞內的複雜度分佈情形。依據視覺複雜度分佈指標、本研究挑選出. 立. 三組不同視覺複雜度分佈的詞彙 (左偏移、右偏移以及無偏移) 、並將這些詞彙箝入. ‧ 國. 學. 於句子中。紀錄中文讀者閱讀這些句子的眼球運動軌跡、以比較三組實驗情境下. ‧. 的初次觸接凝視時間 (first-pass duration) 、落點位置 (landing position) 以及眼跳機. sit. y. Nat. 率指標 (probability measures) 差異。使用線性混合模型 (Linear mixed model, LMM). io. er. 估計實驗組別效果、以探討視覺複雜度分佈指標如何影響決定停留時間與眼跳位. al. 置的機制。結果發現右偏移組落點位置都落在其他兩組的右側、而兩組偏移組的. n. iv n C hengchi U 凝視時間都較不偏移組短。進一步分析顯示上述結果需在眼跳目標為中文詞彙才可觀察到。這指出由視覺複雜度分佈指標所反映出的中文雙字詞明視度型態 (luminance pattern) 、會在該詞彙被凝視之前影響眼跳位置的決定、並依落點位置差異而調節了該詞彙被凝視時的處理。. 關鍵詞：眼跳標靶機制、落點位置、閱讀中文。. viii.

(11) Acknowledgement. This research was done during my affiliation to the Eye Movement and Reading Laboratory (ERMLAB) directed by Professor Tsai Jie-Li ( 蔡介立 ) whose punctual directions and patient tutoring were of great value during my studies at NCCU. The research equipment as well as the members of this laboratory created an excellent environment for my personal and academic growing along this three unforgettable years of my life. Special. 政治大. thanks to Chen Chia-Hsing (陳家興) who supported me along the way with his computation. 立. expertise and statistical knowledge, his patient tutoring, invaluable help and lessons are. ‧ 國. 學. very much appreciated. My most sincere respect and appreciation for Professors Chih-hung Chang (張智宏) and Miao-Hsuan Yen (顏妙璇) who kindly accepted to be members of my. ‧. evaluation committee and whose comments help me to obtain a deeper understanding about. y. Nat. n. al. er. io. this document.. sit. my own ideas through discussion and reflection upon them, thus leading me to complete. Ch. i n U. v. The sum of events and situations that have lead me to complete this document is. engchi. full of learning experiences not only in the academic but also in the personal realm, I feel thankful towards many people who I encountered along this challenging path, people who are worth mentioning due to their unmeasurable support are my parents Amador Cruz, Yaneth Molina and my sister Diana Milena Cruz Molina. Some friends towards who I also feel grateful are Tzu Yun Hung, Julio Noyola, Christopher Young, Manuel Diaz, Robert Wang, Carlos Denegri, Sherry Shieh and Silvia Medina; by the last, my whole gratefulness towards the Professors at the Psychology department of NCCU whose patience and instruction leave indelible marks on my brain’s sulci and fissures. Finally, special thanks to ix.

(12) a woman who believed in me sincerely and doubtless in despite of many difficulties, who give me not only her unconditional support but also her love, Miss Yang.. The forces of evolution have endowed human kind with instruments of a magnificent complexity to contemplate, admire and inquire about an equally complex universe, the eyes.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. x. i n U. v.

(13) 1 Forewords. The act of reading can be apprehended by studying the eye movement pattern along the lines of a text; in a basic level, this pattern can be decomposed in two different ocular behaviors: saccades and fixations. Saccades are small and fast jumps towards different locations in the text, movements that bring new visual information to the foveal region in order to be processed; instead, fixations are the halts between saccades and represent the. 治政大 lexical access and passage duration indicate the ease or difficulty in word identification, 立. time that the eyes remain in one position before the next saccade is executed. Thus a fixation. integration1 (Rayner, Fischer, & Pollatsek, 1998). As reading proceeds, the lexical mapping. ‧ 國. 學. of each word takes place and the overall meaning of the sentence is gradually enriched by. ‧. the information acquired with each fixation2; thus, the interplay between these oculomotor behaviors is informative about ongoing cognitive processes and constitutes the focus of. y. Nat. er. io. sit. research in eye movement studies.. Within this research field a reasonable topic of study concerns about the units of a. n. al. Ch. i n U. v. writing system that are used as the target of saccades in reading, in other words, what are. engchi. the language features upon which readers rely for oculomotor control and saccade programming. The present study attempts to address this question in reading Chinese; for doing it so and due to the particular visual layout of this writing system, an index that. 1. In a reading task using a word masking procedure, participants took about 60 msec to successfully identify. the low-level features of a word, but the eyes remained in the masked location up to 250 msec, time that was thought to reflect a complete semantic retrieval (Rayner & Pollatsek, 1989a). 2. There is a current debate on whether words are activated either sequentially or in parallel, but both points of. view assume that all of the words within a passage are somehow activated during reading for comprehension..

(14) 2 reflects the Visual Complexity Distribution (VCD) within 2-character Chinese words was devised. This index reflects the distribution of visual components along the words’ area, which we consider to be of relative importance in the decisions about where and when to move the eyes in reading Chinese. Previous studies in reading alphabetic languages indicate the importance of lowlevel features in the decision about where to move the eyes next, in particular word length information provided by the blank spaces between words (McConkie, Kerr, Reddix, & Zola,. 治政大languages, first it is necessary features on saccade targeting mechanism in reading different 立 1988; Rayner, 1979). Nevertheless, to estimate the relative weight of linguistic and visual. to acknowledge that different writing systems provide specific cues that might draw the. ‧ 國. 學. native readers’ eyes with differential intensity. The question about what are the traits upon. ‧. which is based the saccade strategy and how consistent is this strategy used by the readers of a certain language, will be discussed in Chapter one of this document into two separate. er. io. sit. y. Nat. sections.. Rayner (1979) was the first author to propose the existence of a saccade strategy in. n. al. Ch. i n U. v. his paper about eye movement guidance models, making reference to the oculomotor. engchi. mechanism in charge of deciding where to send the eyes next; this study is the cornerstone of subsequent researches on the topic and was the first to suggest that the eye movements in reading are made on a non-random basis. This study proposed the existence of a Preferred Viewing Location (PVL) near the word center, result that was further taken as evidence supporting a word-based saccade strategy in reading English, finding that was later replicated in reading other alphabetic languages (Deutsch & Rayner, 1999; Nuthmann.

(15) 3 & Kliegl, 2009; M. Yan et al., 2014); these and other eye movement studies in reading languages other than Chinese will be reviewed in the first section of Chapter one. On the other hand, the existence of a functional target of saccades in reading Chinese is not easily identifiable due to the particular visual layout of this language, though the observation of a PVL-like phenomenon in reading Chinese might suggest that the functional units of saccades remains to be the same across different languages irrespectively of the writing system’s visual layout. The studies in Chinese analyzing the influence of. 治政大 about the computation of the section of Chapter one. This chapter ends with an explanation 立. linguistic and visual factors on saccade targeting mechanism will be reviewed in the second. word’s Visual Complexity Distribution (VCD) index, which we thought to reflects how the. ‧ 國. 學. word’s visual components are scattered within its area and influence the specification of. ‧. saccade targets.. Chapter two describes the experiment designed for assessing the VCD index’s. y. Nat. er. io. sit. influence on saccade targeting mechanism; therefore it includes the experimental question, the working hypotheses, the method employed for selecting the target words and. n. al. Ch. i n U. v. constructing the reading material. This experiment was aimed to estimate to extent to which. engchi. the word’s VCD Index influences the Initial Landing Position (ILP) of the interword saccades, response that is considered to inform about the selection of saccade targets; in addition, analyses on other first-pass eye movement measures are also provided in order to provide a comprehensive picture about the VCD effects. The results were found to be relevant for the discussion on what features of the Chinese script have a role in deciding where to send the eyes next and concurrently validated the VCD index as a measure for.

(16) 4 studying the saccade targeting mechanism and oculomotor control in reading natural sentences. Chapter three includes additional statistical analyses aimed to test a) whether the results presented in Chapter two could be replicated when analyzing a Region of Interest (ROI) that contained a word boundary, thus violating the unity of word entities, and b) whether the VCD influence on saccade targeting could remain when analyzing the ILP within other 2-character words. For doing it so, new ROI with similar VCD indexes than. 治政大about the new ROI selection the measures between different VCD conditions. Details 立. those of the target words were localized within the reading material and then we contrasted. procedure along with the statistical results of these contrasts are further discussed in. ‧ 國. 學. Chapter three.. ‧. Finally, Chapter four consist of the general discussion and organize the findings from previous sections distinguishing the VCD index’s influence on both saccade targeting. y. Nat. er. io. sit. and word identification, sections that represent the acquired understanding about the VCD index’s role in reading Chinese. Chapter four concludes with the exposition of limitations. n. al. Ch. i n U. v. from the present study along with question that can led to future researches. Finally and due. engchi. to the specialized language used along this document, a Glossary of terms is provided at the end with the interest of aiding the reader to interpret the literature, analyses and results presented along this document..

(17) 5. Chapter 1. Literature Review. This chapter presents the findings from previous eye movement studies that addressed the question about what factors influence saccade targeting mechanism in reading tasks. The presentation of findings is distinguished by languages in two sections; the first section reviews the findings in reading languages other than Chinese with a special. 治政 Chinese. Such distinction was motivated by two reasons; 大 first, to expose the relative weight 立. focus on alphabetic languages, instead the second section reviews the studies in reading. of both perceptual and linguistic traits among written systems that greatly vary in terms of. ‧ 國. 學. visual layout, having in one extreme of this continuous alphabetic languages such as. ‧. English and German (i.e. languages that make use of letters to conform words and have blank spaces between words) and in the other extreme logographic languages such as. y. Nat. er. io. sit. Japanese and Chinese (i.e. languages that use logographic units without interword spacing, see Appendix A). The second reason for making this distinction is to provide a common. n. al. Ch. i n U. v. framework in the presentation of findings from both sets of literature that will be useful for. engchi. understanding the methodology and results obtained in the present study..

(18) 6. 1.1. Eye Movement Studies in Reading Languages other than Chinese. 1.1.1. The Influence of Low-level Features on Saccade Targeting. Word Length The decision about where to move the eyes next in alphabetic languages3 appears. 治政大 the Initial Landing Position in particular, two variables have been found to greatly influence 立. to be primarily determined by low-level visual information obtained from parafoveal vision;. (ILP) within words, being those the main cues upon which the oculomotor system relies for. ‧ 國. 學. sending the eyes forward: word length and interword spacing. The results provided by. ‧. Rayner (1979) on the proportion of ILP in English words of different lengths are the cornerstone findings suggesting the influence of low-level features in oculomotor control;. y. Nat. er. io. sit. he found that the ILP within a word is modulated by its length: short words (i.e. ≤ 3 letters) had a higher probability of being initially fixated at the right edge as well as a higher. n. al. Ch. i n U. v. probability of being skipped, while long words (i.e. ≥ 7 letters) were likely to be initially. engchi. fixated at the center and had higher probability of being refixated. That is to say, word length information is used by the English readers for moving their eyes along the text.. 3. The category of alphabetic languages in this context includes those languages that make use of letters (i.e.. vowels and consonants) to conform words/linguistic units that are separated by blank spaces, resulting in words/linguistic units of different length. Such category is not restricted to Indo European languages (i.e. English, German, French, Spanish etc.), but also includes Thai, Hebrew & Uighur, languages which have been also the focus of eye movement studies during the past decades..

(19) 7 Other crucial findings from Rayner’s study were that when making forward saccades the readers’ eyes tended to systematically land between the beginning and the middle of a word as suggested by the distribution of initial fixations, and when making regressive saccades the readers’ eyes tended to land half way between the end and the middle of the word. The distribution of initial fixations was interpreted as an attempt to send the eyes towards the center of the upcoming word, but due to a mixture of influences the ILP was slightly before the intended location 4 . Rayner (1979) interpreted the. 治政 Location (PVL) between the middle and the nearest edge大 of the word, observation that is 立. distribution of initial fixations as evidence about the existence of a Preferred Viewing. regarded as one of the primary indicators that eye movement control in reading English is. ‧ 國. 學. word-based.. ‧. The PVL phenomenon was later replicated in reading German by Nuthmann and Kliegl (2009), who showed that the average fixation positions were close to the word center. y. Nat. er. io. sit. with a leftwards shift with increasing word length for the set of forward single fixation cases. Moreover, a similar analysis in reading Hebrew, a language naturally written from. n. al. Ch. i n U. v. left-to-right, provided qualitatively similar results than those in reading other alphabetic. engchi. languages (Deutsch & Rayner, 1999). These set of results allowed scholars to validate the PVL phenomenon across alphabetic languages; consequently, interword spaces are regarded as salient visual cues readily available from parafoveal vision that are used in deciding where to send the eyes next (see Figure 1).. 4. The influences that explain a shift from the intended target have multiple sources, errors in the oculomotor. program or systematic errors as the saccade range effect (McConkie et al., 1988)..

(20) 8. 立. 政治大. Figure 1. Cross language validation of the PVL phenomenon. ‧ 國. 學. Note: The red points represent the mean ILP of all interword forward saccades for different word lengths. ‧. Thus fixations within alphabetic words of different length is somehow systematic,. sit. y. Nat. as suggested by the observation that readers eyes tended to initially land about halfway. al. er. io. between the beginning and the middle of words (see Table 1). Rayner (1979) labeled this. n. iv n C h eposition by O'Regan (1981), the latter being the h i aUword in which recognition time is n g cwithin. location as the PVL which is different from the Optimal Viewing Position (OVP) described. shorter. Thus the saccade target in reading alphabetic languages is assumed to be the word center which usually corresponds to the OVP, but the eyes normally land short of the OVP onto the PVL due to oculomotor noise in saccade programming and oculomotor execution. In this way, the OVP makes reference to the location in a word where an optimal reading performance is expected and the PVL makes reference to the position within a word where the eyes actually land..

(21) 9 Table 1. Mean ILP by word length and language Word length (letters) Language Author. Frequency –. Rayner (1979). English. 3. 4. 5. 6. 7. 8. 2.1 2.4 2.8 3.2 3.5 3.8. 9. 10. –. –. Rayner et al (1996). LF. –. –. 2.8 3.0 3.7 4.1 4.4 4.7. Rayner et al (1996). HF. –. –. 2.7 3.2 3.6 4.1 4.4 4.5. Hebrew. Deutsch & Rayner (1999). –. German. Kliegl & Underwood (2009). –. Uighur. Yan et al. (2014). 立. 2.3 2.7 2.8 3.2 3.5 3.8. 1.7 2.1 治政 – 2.0 大 2.4. –. 2.5 2.9 3.0 3.2 3.6 3.5 2.7 3.2 3.3 3.8. –. Note: English and German are languages naturally written from left to right; Hebrew and Uighur instead, are. ‧ 國. 學. language that are written from right to left. The column Frequency makes reference to the word frequency; in Rayner et al. (1996) Low Frequency (LF) words had frequencies of 10 per million or less, and High Frequency (HF) words had frequencies of 50 per million or larger.. ‧. sit. y. Nat. Finally, it is worth to mention the correlation between word length and word. io. er. frequency, two word properties that are often used as independent variables in reading studies. In order to assess whether word frequency influences the decision of where to move. n. al. Ch. i n U. v. the eyes next, Rayner, Sereno, and Raney (1996) analyzed the ILP in words of different. engchi. lengths with two levels of frequency (i.e. words with frequencies of 10 per million or less, Low Frequency words, and words with frequencies of 50 per million or greater, High Frequency words). As can be seen in Table 1, the mean ILP for words of the same length varying in frequency was not significantly different from each other, dismissing the role of word frequency on saccade targeting mechanism.. –. –.

(22) 10 Launch Site The selection of a target location within a word can not only be discussed in terms of ILP but also as consequence of the take-off location distance of the immediately previous fixation, referred as the launch site. McConkie et al. (1988) confirmed that the eyes are usually send to a location between the beginning and the middle of a word, they also realized that this position systematically varied as a function of the distance from the previous fixated location up to the target word’s beginning, in such way that for larger. 治政大 the mean ILP shifted rightwards (see Figure 2). 立. launch site distances the mean ILP shifted leftwards while for shorter launch site distances. ‧ 國. y. sit. al. n. 2. er. io. Position (letter). Nat. 3. ‧. 4. 學. 5. 1. Ch. n engchi U. iv. 4-letters 5-letters 6-letters 7-letters 8-letters. 0 -7. -6. -5. -4. -3. -2. -1. Launch site (letter). Figure 2. Mean ILP by word length distinguishing incoming saccades from different launch sites, data from McConkie et al. (1988) Figure 2 shows how the mean ILP shifts rightwards as the launch site gets closer to the target word for words of different length, observations that make of the launch site a.

(23) 11 variable of relative importance when analyzing the distribution of initial fixations and making inferences about the saccade targeting mechanism. Accordingly, saccades send from larger launch sites will result in initial fixations closer to the beginning of words, conversely shorter launch sites will result in initial fixations further into the word. Hence launch site could be arguably considered to be the correlate of an oculomotor constraint in reading, thus indicating that the eyes might have a tendency of making saccades of certain amplitude, distance that matches the perceptual span or the window from which detailed. 治政大 devised by McConkie and As demonstrated by the moving window paradigm 立. information can be successfully obtained.. Rayner (1975), the amount of information that can be successfully retrieved in a given. ‧ 國. 學. fixation is reduced exponentially with eccentricity; in that study the sentence display varied. ‧. following the reader’s oculomotor pattern in such a way that some of the letters within a sentence were masked based on the current position of the eyes. Thus and for a given. y. Nat. er. io. sit. fixation, only certain parts of the sentence were unmasked or available for the reader. Through systematic variation in the amount of available text to the left and right of the. n. al. Ch. i n U. v. currently fixated point, it was possible to establish that word length patterns can be acquired. engchi. further into the peripheral vision up to approximately 3° of Visual Angle (VA), and that oculomotor guidance was not influenced by information beyond 12 letter positions. The previous observations redefined the PVL as the maximum point in the distribution of all the initial fixations within a word, being a composite distribution in nature. McConkie et al. (1988) noticed that given a constant launch site and word length the ILP distribution in a word was normally distributed and from this acknowledgment, scholars redefined the PVL as a launch site contingent distribution. Accordingly, the PVL curve is.

(24) 12 a weighted average across several launch site contingent distributions, with word length only having a small effect on the means and variability of the ILP distribution. These findings imply a tendency to send the eyes towards a center of gravity influenced by the luminance patterns and constrained to a normal distribution within a word (Tsai, 2014; Yang & McConkie, 1999).. Lack of Interword Spaces. 治政大then use it for selecting the next boundary information retrieved from parafoveal vision and 立 One implication of a word-based saccade strategy is that readers use the word. saccade target (Rayner et al., 1996). In English as well as in other alphabetic languages, the. ‧ 國. 學. knowledge about the beginning and end of words is provided by interword blank spaces,. ‧. but to what extent is this cue necessary for efficient oculomotor guidance? In order to assess the effect of omitting word boundary information in the selection of saccade targets Rayner. y. Nat. er. io. sit. et al. (1998) presented sentence in which the blank spaces were either filled with x or removed (see Figure 3). Under these unusual reading situations, the reading rate decreased. n. al. Ch. i n U. v. nearly 50% due to an increase in fixation durations, a shortening in the length of forward. engchi. saccades and an increase in the number of regressions; additionally, the PVL curve shifted towards the word beginning and become a negative linear slope. Therefore semantic boundaries indicated by interword spaces were found to play an important role in deciding where to send the eyes next in reading alphabetic languages. The omission of blank spaces from the written script greatly disrupted the reading performance, increasing word identification times and shortening saccade lengths; in other words, both temporal and spatial characteristics of the eyes’ behavior were found to be affected by the.

(25) 13 omission of interword spaces (Rayner et al., 1998). Therefore, interword spaces have a special status in reading alphabetic languages, being salient cues that are used to isolate words parafoveally, information that is used for the readers to move their eyes along the text with a role in the specification of saccade targets. The PVL curves for words in unspaced conditions in Rayner et al. (1998) no longer followed a Gaussian shape but turned into a negative linear slope decreasing from the word beginning, finding that was later replicated by Kanjii, Nazir, and Osaka (2001) in reading. 治政 the simultaneous use of two syllabic scripts or Kana (i.e. 大 Hiragana and Katakana), used to 立 Japanese. Japanese is a language naturally written without interword spaces that includes. write particles, auxiliary verbs and grammatical elements, and loaned Chinese characters. ‧ 國. 學. or Kanji, used to write nouns, verbs, adjectives and some adverbs (Trigger, 2004).. ‧. The study of Kanjii et al. (2001) suggested that in reading languages without clear interword spaces, readers might use a different set of visual cues to aid the ongoing. y. Nat. er. io. sit. language processing and the saccade targeting mechanism; in order to explain this negative linear slope without disregarding the role of semantic units in oculomotor control, they. n. al. Ch. i n U. v. suggested that the simultaneous use of Kana and Kanji characters provide word. engchi. segmentation cues that can be effectively used by Japanese readers to segment words parafoveally. This is so because the beginning of Japanese words is usually occupied by Kanji characters, which are visually more salient than other grammatical components written in Kana characters, low-level feature that was argued to add attractiveness to the word beginning to be selected as the saccade target. A subsequent study testing the effect of inserting blank spaces in Japanese gave support to the previous remarks. Sainio, Hyönä, Bingushi, and Bertram (2007) asked native.

(26) 14 Japanese speakers to read Pure Hiragana or mixed Kanji – Hiragana sentences with and without interword spaces (see Figure 3). When reading Pure Hiragana text eye guidance was facilitated by the insertion of spaces, but when reading mixed Kanji – Hiragana text the PVL at the word beginning was not significantly affected by the insertion of spaces. These findings suggest that interword spacing in reading natural Japanese provide redundant information because the visually salient Kanji characters, which predominantly appear at the beginning of words are effective segmentation cues per se.. 政治大. Normal spacing. 立. This is a sample sentence. No space. ‧ 國. 學. Thisisasamplesentence. Filled space. ‧. Thisxisxaxsamplexsentence.. y. Nat. くうこうの. io. Unspaced Hiragana. けいえいはげんじょうではおせじにもゆうりょうとはいえない。. sit. なりた. al. er. Spaced Hiragana. n. なりたくうこうのけいえいはげんじょうではおせじにもゆうりょうとはいえない。 Spaced Kanji – Hiragana. 成田空港の. 経営は. Ch. engchi. 現状ではお世辞にも. i n U. v. 優良とはいえない。. Unspaced Kanji – Hiragana. 成田空港の経営は現状ではお世辞にも優良とはいえない。. Figure 3. Spacing conditions in Rayner et al. (1998) and Sainio et al. (2007) Note: The English translation for the Japanese sentence is “The management of Narita airport can by no means said to be excellent”. On a different account the analyses on ILP in reading spaced and unspaced Thai, an language that is naturally written without inter word boundaries, evidenced that spacing.

(27) 15 neither facilitated nor disrupted saccade targeting or early lexical segmentation and the ILP distribution for both conditions peaked at the left from the word center (Winskel, Radach, & Luksaneeyanawin, 2009). The similar patterns of ILP distribution for both spacing conditions indicated that parafoveal word segmentation in reading Thai might rely on information other than spaces. One answer that supports this suggestion, came from diagnosing the probability that letters had of occupying the word’s beginning or end within the Thai lexicon; according to. 治政大was equally high for a set of 10 accounted for a 77% of all word ending letter positions and 立. the analysis performed by Reilly, Radach, Corbic and Luksaneeyanawin (2005), 10 letters. letters to occupy the word beginnings (54%). Given the probability of certain letters to. ‧ 國. 學. appear either at the beginning or at the end of words, it is likely that native Thai readers. ‧. become aware of it and use this information as word boundary cues in the absence of clear spaces between words.. y. Nat. io. sit. Thus, the PVL in languages naturally written without spaces, such as Japanese and. n. al. er. Thai, was found to be at the word’s beginning, indicating that the mechanism underlying. Ch. i n U. v. parafoveal specification of saccade targets might rely on traits of diverse nature in reading. engchi. different writing systems. Given these observations a word-based strategy might not universally rely on the presence of blank spaces, instead is likely to be constrained by the visual layout and the familiarity of native readers with specific word segmentation cues of their language, leaving open to research the spectrum of traits that can be effectively used to guide the eye movements in reading different writing systems..

(28) 16. 1.1.2. The Influence of High-level Features on Saccade Targeting. Morphological Traits Most of the studies in reading alphabetic languages support the idea that deciding about where to move the eyes next extensively relies on low-level visual information and other studies testing the influence of high-level features seem to validate this statement.. 治政 guidance in reading French; for doing it so she selected 大 suffixed and prefixed words with 立 Beauvillain (1996) tested whether morphological information is used for oculomotor. two levels of frequency matched in length as target words5, the analysis on ILP responses. ‧ 國. 學. did not show significant differences between suffixed and prefixed words when word length. ‧. was matched, being the 3rd or 4th letter position the more frequent initial landing site. Similarly, the study conducted by Deutsch and Rayner (1999) provided evidence of. y. Nat. er. io. sit. the reduced influence of Hebrew words’ morphological structure on ILP responses; in this study the PVL was right to the center of words without significant difference between single. n. al. Ch. i n U. v. and plural Hebrew words varying in length. However the 2nd and 3rd experiment in this. engchi. study showed that words’ morphology does modulate the OVP. Hebrew is a language written from right to left and accordingly the reading direction modified the position of the PVL, nonetheless the results were qualitatively similar to the findings in reading English. The experimental manipulation of morphological aspects as in Deutsch and Rayner (1999), is part of a group of studies testing the influence of affix forms on saccade targeting. 5. An example of a Suffixed word pair with different frequency is tempérament (freq. = 979) and combinateur. (freq. = 3), and a Prefixed word pair with different frequency is transformé (freq. = 692) and entreposer (freq. = 15). These examples are extracted from the Appendix in Beauvillain (1996)..

(29) 17 mechanism; for example, Inhoff, Briihl, and Schwartz (1996) tested the effect of three types of target words whose lexical information was distributed differentially. The target words in this study were 9-letter length noun words consisting of bimorphemic compound words (e.g. blueberry; meaning defined by its ending), bimorphemic suffixed words (e.g. ceaseless; meaning defined by its beginning) and monomorphemic controls (e.g. arthritis) embedded within neutral sentence context. According to their findings, the mean ILP was closer to the word center for. 治政大were attributed to the influence saccades were larger towards compound words, results that 立. bimorphemic compounds than for the other two conditions, thus reflecting that inter word. of their differential distribution of lexical information. Studies alike are aimed to test up to. ‧ 國. 學. what extent high-level linguistic information obtained parafoveally can guide oculomotor. ‧. programming but only a handful of results indicate a significant influence of these traits on saccade targeting (see Table 2).. y. Nat. er. io. sit. The study of Yan et al. (2014) in reading Uighur tested the effect of morphological complexity on ILP. As Hebrew, the natural reading direction of Uighur is from right to left. n. al. Ch. i n U. v. and the PVL analyses for words of different length replicated the results in Deutsch and. engchi. Rayner (1999) and are qualitatively comparable to the findings in reading other alphabetic languages (see Table 1). Additionally, they reported a significant effect of morphological complexity on the selection of saccade targets; mainly, the PVL shifted towards the word beginning with increased number of suffixes as well as for words with high frequency..

(30) 18. Table 2. Mean ILP in different studies testing the morphological structure of words Word length (letters) 7. 8. 9. Compound. –. –. –. –. 4.4. Suffixed. –. –. –. –. 3.9. Monomorphemic. –. –. –. –. 3.7. –. –. –. 3.8. –. –. –. 3.8. 3.2. 3.6. 3.8. –. 3.3. 3.5. 3.8. –. Suffixed 治 – 政大– Prefixed. Beauvillain (1996). 立. Singular. 2.8. Plural. 2.7. Deutsch & Rayner (1999). io. al. ‧. 0 Suffix. –. 3.4. 3.5. 3.9. 4.1. 1 Suffix. –. 3.1. 3.2. 3.9. 4.1. 2 Suffixes. –. 3.4. 3.4. 4.2. y. Yan et al. (2014). Nat. Uighur. 6. 學. Hebrew. Inhoff et al. (1996). 5. sit. French. Condition. 3.0. er. English. Author. ‧ 國. Language. v. n. Note: English and French are languages naturally written from left to right; Hebrew and Uighur instead, are naturally written from right to left.. Ch. engchi. i n U. From these results Yan et al. (2014) concluded that Uighur readers might be able to decompose the word about to be fixated into morphological units in their parafoveal vision and gave greater importance to the root morpheme, selecting this area as the saccade target rather than towards the suffixes. They also suggested that as for evidencing a saccade targeting effect the eyes are required to be within at least 10 letters from the word beginning, result that is comparable in degrees of VA as the distance from which information influences oculomotor control in reading English (McConkie & Rayner, 1975)..

(31) 19 Additionally, Yan et al. (2014) performed an analysis for estimating to what extent the readers of Uighur used morphological information to select the next saccade target; the results suggest that there might be different reading strategies or skill levels within the tested subjects, this is so because subjects who made longer interword saccades exhibited an overall larger word length effect and a smaller morphological complexity effect. In addition, subjects who showed a positive morphological complexity effect simultaneously showed a positive word frequency effect, but were not that sensitive to the word length information, as indicated by a negative word length effect. Finally, they didn’t find any. 治政大 and pre–) versus inflectional differential influence of derivational (e.g. like English –ness 立. (e.g. like English past tense –ed or plural –s) suffixed words on ILP.. ‧ 國. 學. As can be seen there is a mixed pattern of results concerning the influence of. ‧. morphological structure of words on saccade targeting mechanism, but a shift of the ILP further into the word for compound words in Inhoff et al. (1996) and for words without. y. Nat. io. sit. suffixes in Yan et al. (2014) is likely to be caused by a differential processing of the. n. al. er. upcoming word’s morphological subunits as reflected by a mean ILP further into the word. Ch. i n U. v. for those conditions. It is also possible that those differences were caused by easiness in the. engchi. parafoveal processing of such conditions due to differential letter transition frequencies at the morphological boundary or be attributed to individual differences.. Word Predictability Lavigne, Vitu, and d’Ydewalle (2000) reported a rightwards shift of the ILP distribution for predictable and highly frequent target words when prior fixations were close to the beginning of the target word, suggesting that ongoing perceptual and linguistic.

(32) 20 processes can affect the selection of saccade targets. This finding was explained as consequence of an ease in processing which enhanced the channel capacity that retrieves information parafoveally. This study as most of the previous studies on the influence of high-level features, stressed the effect of launch site in the saccade targeting mechanism; this is, the closer the eyes are from the beginning of the upcoming/target word, there is larger opportunity to retrieve information from that word and when the last fixation is beyond 12 letter positions from the Region of Interest (ROI), parafoveal processing from. 政治大. the ROI is unlikely to influence oculomotor guidance.. 立. Eye Movement Studies in Reading Chinese. 1.2.1. The Chinese script. ‧. ‧ 國. 學. 1.2. Nat. sit. y. Before dealing with the issue of saccade targeting in reading Chinese it is necessary. n. al. er. io. to briefly review some features of this language, which make of it a unique scenario to. i n U. v. investigate eye movement and the associated cognitive processes. As no other modern. Ch. engchi. language in the world, Chinese is a pure logographic language6 that uses characters as its writing unit; the visual arrangement and number of strokes within the characters’ box-like shape area varies greatly resulting in more than 5000 characters, each one associated with. 6. According to Robertson (2004) a logographic language can be either a language conformed by a) units that. have an iconic relationship, showing a similarity between the object and the written sign (i.e. Pictographic relationship); or by b) units that have a symbolic relationship, indicating an association of habit between the written sign and the object of the spoken sign (i.e. Ideographic relationship). In early Chinese the logographs were realistic pictures which stood for the name of things, but later these logographs changed from pictographs to ideographs, where the association become habitual and was no longer based on resemblance..

(33) 21 a syllable and a tone. Nowadays the reading direction of Chinese is left to right, but it is also traditionally written vertically from top to bottom in advertisements, newspaper and other publications. Characters are the writing unit of Chinese language but single character words account for approximately 15% of the words, while 2-character words account for more than 60% of the words within the Chinese lexicon. Due to the lack of interword spaces the word concept is not as clear as in spaced alphabetic languages, in which the definition relies. 治政 visually salient objects. In alphabetic languages interword大 spaces are low-level visual cues 立. on this orthographic feature that clearly separates strings of letters and make of words. that guide oculomotor control and aid the visual system in the selection of saccade targets;. ‧ 國. 學. in comparison and due to the visual layout of Chinese, there is a debate about the unit upon. complex characters without clear semantic boundaries7.. Nat. y. ‧. which is based the native readers’ saccade strategy, that involves parsing a string of visually. er. io. sit. Another unique characteristic of Chinese is that there are two variations of its characters; the traditional characters (i.e. 繁體字) in contrast to the simplified characters. al. n. iv n C (i.e. 简体字), have a relatively higher of visual complexity and the history of some h elevel ngchi U. traditional characters can be traced up to the 5th century AD. Instead, the simplified Chinese contains newly created characters or radical substitutions performed from the 50’s by the People’s Republic of China government, which substantially reduced the character’s visual complexity in order to increase the levels of literacy among the military and civilian population. Traditional characters are mainly used in Taiwan, Hong Kong and Macau while. 7. See Appendix A for a comprehensive comparison of visual, linguistic and phonetic traits between Chinese. and alphabetic languages..

(34) 22 simplified characters are used in the Mainland China (Yan, Kliegl, Richter, Nuthmann, & Shu, 2010; 何沐容, 2010). As in the previous section of this Chapter, the influences of low-level features on saccade targeting will be presented in the first place, followed by the influences of highlevel features; although this division is useful for explanatory purposes, the greater part of studies manipulate a combination of these traits, therefore it is recommended to integrate the findings between separated sections. The last part of this Chapter is committed to. 政治大 index of Chinese words as well as its implications for previous studies. 立. explain the procedure followed for computing the Visual Complexity Distribution (VCD). ‧ 國. 學. 1.2.2. The Influence of Low-level Features on Saccade Targeting. ‧ sit. y. Nat. Word Length. io. er. Some of the studies in reading Chinese have dealt with the question about what information is readily available in parafoveal vision for guiding the eyes towards the next. al. n. iv n C landing location, being word length the first feature that U h e n g c h i attracted the scholars’ attention; in. particular, they are interested in knowing whether the ILP distribution within words of different length resembled the PVL in reading alphabetic languages, bearing in mind the lack of interword spaces added to the particular visual layout of the Chinese script. The first study that reported a PVL analysis in reading Chinese was made by Yang. and McConkie (1999), their results on the proportion of eye fixations that initially landed at different position within 2-character words did not support the existence of an identifiable PVL. In this study the distribution of initial fixations did not peak at any particular area.

(35) 23 within the word and all of the areas in which the ROI was divided received a similar amount of initial fixations (see Figure 4). This result was later replicated by Tsai and McConkie (2003), study in which the PVL curves obtained for single characters and 2-character words were flat. These observations indicated that words of different length do not function as saccade targets in Chinese as they do in alphabetic languages, even if has been recognized their psychological reality within the Chinese reader lexicon (Bai, Yan, Zang, Liversedge, & Rayner, 2008;. 治政 Yen, Radach, Tzeng, and Tsai (2012) tested the 大 degree of word n + 1 parafoveal 立. Inhoff & Liu, 1998; Shu, Zhou, Yan, & Kliegl, 2011).. preprocessing and its relationship with the saccade targeting mechanism; for doing it so,. ‧ 國. 學. they used the invisible boundary paradigm and manipulated the preview time of the. ‧. upcoming word. Their findings suggest that parafoveal information of the upcoming word was extracted relatively later during ongoing fixations as reflected by a leftwards shift of. y. Nat. er. io. sit. the PVL when the parafoveal region was masked at the end of the previous fixation. They suggested that parafoveal information at least at the character level is extracted relatively. n. al. Ch. i n U. v. early, giving the constraints that Chinese writing system impose to the readers, namely a. engchi. large variety of visual patterns along the string of character and the lack of salient word boundaries. Li, Liu, and Rayner (2011) compared the PVL curves of 2 and 4-character words embedded in identical sentence frames; in this study the proportion of initial fixations landing at different character position along a 4-character ROI was not significantly different for words varying in length, with the ILP distribution peaking at the word beginning for both conditions. This finding suggest that saccades are not sent towards word.

(36) 24 units, probably because the lack of interword boundaries decrease the possibility of parafoveal word segmentation. Li et al. (2011) argued that if words function as the unit upon which saccade targeting is based, their results would have indicated a rightwards shift of the PVL when the eyes are sent towards 4-character words in comparison to 2-character words. In contrast to these results, the ILP curves in Yan, Kliegl, Richter, et al. (2010) evidenced a rightwards shift of the PVL with increased word length, being the sole study evidencing this pattern (see Figure 4). Overall these findings do not support a word-based. 治政大the eyes next due to the visual available to be used as a cue to for deciding where to move 立. saccade strategy in reading Chinese, or at least that word length information is not readily. layout of Chinese8.. ‧ 國. 學. These results indicate that the visual layout of Chinese might difficult the. ‧. information retrieval about the length of word n + 1, nevertheless the study conducted by Wei, Li, and Pollatsek (2013) found that properties of the currently fixated words such as. y. Nat. er. io. sit. word length, do influence oculomotor guidance as demonstrated by the average progressive outgoing saccade length. Saccade amplitude was found to be relatively longer when the. n. al. Ch. i n U. v. eyes left a 4-character word in contrast to the observations when the eyes left a ROI with. engchi. two 2-character words. Thus while parsing a Chinese sentence, a top-bottom process with different automation values derived from the ease of lexical identification of the semantic unit at the currently fixated location, might influence oculomotor control as well as the extend in which parafoveal information is used to select the next saccade target.. 8. This can be accounted by three embedded factors of Chinese characters: a) visual acuity, due to a larger. visual angle of the characters in comparison with the letters of alphabetic languages, b) meaning compactness, the information contained in each character is higher than that of individual letters, and c) visual complexity, characters vary in number of strokes and radicals composition..

(37) 25. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 4. PVL curves from previous studies in reading Chinese. 25. Note: Proportion of eye fixations that initially land at different positions within Chinese words of different length.

(38) 26 From this perspective Chinese readers are thought to use their lexical knowledge to segment the characters into words, process that might be relatively more demanding than relying exclusively on interword spaces, which simultaneously diminishes the influence that word n + 1 length information could have in deciding where to send the eyes next. However and in the quest for demonstrating an identifiable PVL in reading Chinese, the analyses on ILP responses had become more detailed. Yan, et al. (2010) considered separately two sets of ILP responses: a) single. 治政 some evidence about a PVL-like phenomenon in reading 大 Chinese. The proportion of single 立. fixations and b) the 1st of two fixations cases, this method of data segmentation provided. fixation cases resembled the results in reading alphabetic languages, with an identifiable. ‧ 國. 學. peak near the word center as indicated by the mean ILP; in contrast, the distribution of. Table 3).. ‧. initial fixations from the 1st of multiple fixations cases peaked at the word beginning (see. y. Nat. er. io. sit. Finally, Li and Shen (2013) tested the joint influence of word length and interword spaces in the selection of saccade targets by inserting blank spaces before and after 2- and. n. al. Ch. i n U. v. 4-characters target words embedded in the same sentence frames. The ILP distribution. engchi. analyses in this study did not provide evidence about the Chinese readers sending their eyes more frequently towards the word center when a space was inserted either before or after target words of different length, but there was a higher probability of sending the eyes further into the word when a space was inserted before. The findings in Li and Shen (2013) as well as those in Yan et al. (2010) were used to propose that Chinese readers dynamically select the next landing position depending on failure or success when segmenting the words in parafoveal vision, but that word length.

(39) 27 information might not be readily available in reading naturally unspaced Chinese unless the previous fixation is close enough to the upcoming words as to identify a semantic boundary. In addition, when a clear boundary is available in the parafoveal vision, native readers are flexible enough as for changing their saccade strategy and send their eyes to a position that makes an optimal use of the perceptual span9 amplitude, thus avoiding to land closer to the interword spaces when those provide reliable information about word boundary.. 治政 The insertion of spaces is among the experimental大 manipulations that were carried 立 Insertion of Interword Spaces. out to find a clear PVL in reading Chinese because in comparison to alphabetic languages. ‧ 國. 學. Chinese is a language naturally written without spaces, this trait sometimes leads to. ‧. disagreements about the position of word boundaries in natural sentences (Bai et al., 2008). Up to some extend characters can be regarded as complete linguistic units 10, but some. y. Nat. er. io. sit. characters are bounded to be synchronically used and are meaningless standing alone (e.g. 萄, part of other words as 葡萄, meaning grapes; and 蜘 in 蜘蛛, meaning spider);. al. n. iv n C however, single character words only account for 15%U h e n g c h i of the words within the Chinese lexicon11 (Ching, 2007).. 9. The perceptual span is the area of effective vision in reading and extends one character to the left and up to. 2 ½ characters to the right in reading Chinese (Inhoff & Liu, 1998). 10. Chinese characters are the writing units of the spoken language and simultaneously convey phonetic and. semantic information, thus mapping onto syllables and morphemes (Rayner, Li, Juhasz, & Yan, 2005). 11. The percentage of words by length will ultimately depend on the word set taken for its computation; the. word length distribution from the Academia Sinica Balanced Corpus (2003) evidenced that the proportion of words conformed by 1-, 2-, 3-, 4- and 5- or more character is: 2.98%, 51.15%, 34.08%, 9.28% and 2.48%, respectively..

(40) 28 Visually, characters are well integrated units that occupy a regular size and are separated from each other; therefore, has been said that their box-like shape with their perceptual outlines might exploit more efficiently the circular fovea than English words. These visual and linguistic features led scholars to propose Chinese characters as perceptually-favored saccade targets (Sun, Morita, & Stark, 1985; Tsai & McConkie, 2003). In addition, the lack of interword spaces added to the characters’ visual complexity might demand comparatively more cognitive resources from the reader, and both of these. 治政大readers (Sun et al., 1985). The readers, who made in average shorter saccades than English 立. traits have been demonstrated to have an overall impact on the saccade length of Chinese. comparatively shorter saccade length of Chinese readers has been also interpreted as. ‧ 國. 學. consequence of a smaller perceptual span due to the higher information density of Chinese. ‧. characters (Hoosain, 1991). In view of this and according to Li, Rayner, and Cave (2009), Chinese readers might rely on word knowledge to segment the string of characters into. y. Nat. er. io. sit. words, because the lack of clear interword spaces might decrease the possibility of parafoveal word segmentation and constraint the plausibility of a word-based saccade. n. al. strategy.. Ch. engchi. i n U. v. Bai et al. (2008) tested the effects of placing interword spaces in reading Chinese, their experiment demonstrated that this layout was as easy to read as the visually familiar unspaced text, as reflected by analyses on global reading performance measures; nonetheless, word boundary demarcation facilitated word identification at some degree as reflected by the local duration measures. They also tested the effect of two additional layouts; a non-word spacing condition and a space between every character condition,.

(41) 29 leading to even longer reading times owing to incongruent linguistic information, finding that corroborates the psychological reality of words in Chinese (see Figure 5).. Normal unspaced condition. 科学技术的飞速发展给社会带来了巨大的变化。 Single Chinese character spaced condition. 科. 学. 技. 术. 的. 飞. 速. 发. 展. 发展. 给. 社会. 给. 社. 会. 带. 来. 了. 巨. 带来. 了. 巨大. 的. 变化。. 政社治会带来了巨大. 大的. 变. 化。. 大. 的. Word spaced condition. 科学. 技术. 的. 飞速. Non-word spaced condition. 科. 学技. 术的飞. 速发. 立. 展给. ‧ 國. 學. Figure 5. Spacing conditions used in Bai et al. (2008) and Shen et al. (2012) Note: The English translation for the sentence is “The rapid development of science and technology brings. ‧. great changes to society”.. y. Nat. sit. Bai et al. (2008) did not provide any analysis on ILP distribution, but from the. n. al. er. io. results provided by Li and Shen (2013) who followed a similar manipulation, can be. i n U. v. inferred that when placing a space before a word the PVL is likely to shift approximately. Ch. engchi. one character rightwards, to a position that optimally fits the perceptual span acuity boundaries. It is worth mentioning that the perceptual span of Chinese readers is asymmetric, extending one character to the left and up to 2 ½ characters to the right from the currently fixated position12, covering an area of approximately 3° of VA (Inhoff & Liu, 1998). Thus, the insertion of interword spaces had an influence on saccade targeting. 12. Sun et al. (1985) pointed out that only the region within about 2° of VA around the foveated area can be. used for identifying linguistic or visual patterns, when one character occupy approximately 1° of VA.. 变化。.

(42) 30 mechanism; increasing the likelihood of landing at the word center when the eyes were sent towards 2-character words. The study of Shen et al. (2012) tested the effects of inserting interword boundaries for non-native readers when reading Chinese (i.e. English, Korean, Thai and Japanese native speakers). In this study was assumed that the insertion of spaces would facilitate word parsing and comprehension as a consequence of providing salient saccade targets, which in turn could aid parafoveal word segmentation before they were engaged in lexical. 治政大 attention (see Figure 5). They also hypothesized that commonalities between languages, 立. identification, allowing readers to target their saccades more efficiently and to focus their. such as the lack of interword spaces (i.e. Chinese-Thai) or the use of logographic characters. ‧ 國. 學. (i.e. Chinese-Japanese), could facilitate reading performance; thus extensive experience. ‧. reading a spaced language, for example, could increase the dependency on interword spaces for oculomotor guidance.. y. Nat. er. io. sit. Shen et al. (2012) found that the insertion of interword spaces in contrast to the normal condition, aided the non-native readers performance as demonstrated by the. n. al. Ch. i n U. v. analyses of their global and local eye movement measures, irrespective if their native. engchi. languages provided word boundaries or not. Additionally, the influence of non-word spacing and character-spaced conditions on eye movements was found to be more disruptive with increased Chinese proficiency; consequently, more fluent readers experienced greater disruption and were more sensitive to the lexical status of words. Their study also showed that more fluent readers were more likely to send their saccades closer to the word center, which in turn reduced early identification times and increased overall reading performance, but similar results were found in both spaced.

(43) 31 conditions. The PVL analysis also showed that non-native readers targeted their saccades similarly under both conditions even when analyzing separately single fixation cases and 1st of multiple fixation cases; for the set of single fixation cases the ILP responses were distributed about the PVL, but for the 1st of multiple fixations the distribution peaked at the beginning of words and decreased linearly. The results for the 1st of multiple fixations were explained by a need to refixate a word when the initial fixation was far away from the word center and also has been said to. 治政 resources, b) the transition probability between characters大 at the word boundary and c) the 立. be consequence of other multiple factors such as a) the current allocation of attention. word expectation from the previous context (Yan, Kliegl, Shu, Pan, & Zhou, 2010). On the. ‧ 國. 學. contrary, single fixations are likely to be cases in which the eyes land closer to or at the. ‧. PVL so one fixation suffices; consequently, word identification times are reduced and there is no need to get a second visual sample of the currently fixated word (Yang, 1997).. y. Nat. er. io. sit. These arguments give support to Zang, Liang, Bai, Yan and Liversedge (2013) findings, who compared the reading strategy used by children and adults; in this study,. n. al. Ch. i n U. v. children adopted a more restrained targeting strategy reflecting their less developed reading st. engchi. skill, sending their eyes to the 1 character when reading spaced text and with an increased likelihood of refixating words when their ILP was closer to the left boundary. Instead, adult readers’ saccade strategy was more flexible, sending their eyes towards the upcoming two characters in the unspaced condition and towards the word center in the spaced condition, as consequence of their comparatively greater ability to decompose visually complex patterns (see Table 3)..