Preamble

Running title: Aging Changes the Word-Length Effects in Chinese Reading: Insights from Co-registering EEG Data to Eye Movements

Authors: Yujiao Wen¹, Shiman Hu¹, Zhifang Liu¹*, Zhijun Zhang²
(1. Department of Psychology, Hangzhou Normal University, Hangzhou, China, 311121; 2. Department of Psychology and Behavioral Sciences, Zhejiang University, Hangzhou, China, 310058)

Correspondence: Zhifang Liu, Department of Psychology, Hangzhou Normal University, 2318 Yuhangtang Road, Hangzhou City, Zhejiang Province, China. Email: liuzhifang@hznu.edu.cn, Telephone: +8613738046883

Abstract

Background: Previous research has observed larger word length effects in Chinese reading, speculating that older adults may adopt a character-based processing strategy for recognizing long words. However, to our knowledge, no direct evidence has supported this hypothesis.

Methods: By co-registering EEG data with eye tracking, we investigated how aging affects word length effects in natural Chinese reading.

Results: Our results revealed larger word-length effects on fixation time measures among older adults compared to younger adults. Regarding EEG data, we observed inverse word-length effects on the earliest components and later N400, with word length decreasing amplitudes for older adults. No age-related differences in the word-length effect were observed in the P600 component.

Conclusions: Overall, our findings suggest that older adults adopt a character-based processing strategy for recognizing long four-character words, while employing a similar mechanism to younger adults for integrating words during reading comprehension.

Keywords: older adult reading; word-length effects; Chinese reading; fixation-related potentials

Introduction

Word length is an essential parameter that influences lexical decoding in reading, with short words receiving fewer fixations, shorter reading times, and being skipped less frequently than long words (Gollan et al., 2011; Kliegl et al., 2004; Rayner et al., 2011). Evidence indicates that word length effects in alphabetic script reading are preserved among older adults. For example, von der Malsburg et al. (2015) found no reliable age-related changes in word-length effects on scan paths. Additionally, the distributions of fixation landing positions on long words were comparable across younger and older adult readers (Pollatsek et al., 2006), and fixation location effects on both short and long English word identification were well preserved with aging (Li et al., 2017). Empirical evidence also shows that removing interword spaces disproportionately impairs reading performance in older adults compared to younger adults, highlighting the essential role of word spacing in preserving word length and fixation landing position effects during aging (McGowan et al., 2014, 2015; Rayner et al., 2013).

It is valuable to examine whether these preserved word length effects among older adult readers in alphabetic scripts generalize to non-alphabetic scripts such as Chinese, which has many unique characteristics. In Chinese writing systems, the basic orthographic unit consists of square-shaped characters, and word boundaries are not demarcated by spaces. Most words are composed of multiple characters, with each character having its own semantic meaning that may contribute to the whole word meaning. Despite rigorous evidence supporting the importance of word recognition for text comprehension (Bai et al., 2008), characters also serve as processing units that influence eye movements (Chen et al., 2003; X. Li et al., 2014; Yan et al., 2006). Li and Pollatsek (2020) and colleagues proposed a model in which word decoding involves processing the whole word, its constituent characters, and their interactions (X. Li & Pollatsek, 2020). Investigating Chinese word length effects and their modulation by aging is beneficial for understanding the word and character processing mechanisms that underpin reading.

One important justification for investigating aging impacts on word processing in Chinese reading is that older Chinese readers employ different oculomotor strategies compared to older readers of alphabetic scripts. Specifically, older adults exhibit increased fixation frequency and duration, more regressions, longer progressive saccades, and greater word-skipping behavior compared to younger adults during English text reading (McGowan et al., 2014; Pollatsek et al., 2006; Rayner et al., 2009), suggesting they may adopt a risky strategy to compensate for declines in cognitive processing (Pollatsek et al., 2006; Rayner et al., 2011). In contrast, Chinese older adults make more and longer fixations but are less likely to skip words, suggesting they employ a more careful strategy (Liu et al., 2017; Wang et al., 2018). Research on this topic contributes to understanding both universal processing mechanisms underlying reading and unique processing mechanisms of the Chinese language.

Contrary to patterns observed in alphabetic scripts (e.g., English), Chinese older adults do not demonstrate preserved word-length effects. Specifically, Li et al. (2018) observed that older adults showed larger word-length effects on fixation measures than young adults, while landing their saccades less frequently at optimal intraword locations and generating more refixations on long words during reading (S. Li et al., 2018). Additionally, fixation landing position distribution and its impact on word recognition were not preserved for Chinese older adults. Liu et al. (2015) found that the optimal viewing position effect in Chinese word identification was absent among older adults, who demonstrated peak recognition efficiency when initial fixations were positioned on the first character across all word lengths (two-, three-, and four-character words; P. Liu et al., 2015). These researchers argued that older adults may adopt character-based strategies to comprehend Chinese texts (S. Li et al., 2018; P. Liu et al., 2015).

In the present study, we aimed to establish whether older Chinese adults use a different character-based mechanism compared to their younger counterparts for producing word length effects. There are compelling reasons to suppose they use character-based mechanisms. First, declines in parafoveal vision, smaller perceptual span, and reduced parafoveal processing have been observed for older adults (He et al., 2021; Owsley, 2011; Zhang et al., 2019), which might impact target saccades and lead them to expend more effort decoding long words. Second, the lack of optimal viewing position effects for older adults suggests they adopt a character-based strategy to decode words (S. Li et al., 2018; P. Liu et al., 2015). Nevertheless, there is also evidence that older adults may use similar word-based mechanisms for producing larger word length effects. For instance, an eye-tracking study revealed that older and young adults share similar effects of contextual plausibility on word segmentation (L. Li et al., 2024).

Eye-tracking measures merely reflect the final outcomes of information processing and are thus insufficient for elucidating the cognitive mechanisms underlying the enhanced word length effects observed in older adults. Larger word-length effects could be interpreted as older adults using character-based strategies for reading (S. Li et al., 2018; P. Liu, 2015) or simply needing more effort to decode long words. Event-related potentials (ERP) provide continuous measures of neurocognitive processing and are well-suited for addressing this issue. However, most ERP studies to date have focused on how young adults exhibit word-length effects in alphabetic script reading. For example, one ERP study using rapid serial visual presentation (RSVP) demonstrated augmented word-length effects manifested as early negativity (peaking around 200 ms) distributed across occipital and parietal electrode sites (Van Petten & Kutas, 1990). Subsequent lexical decision studies revealed that word-length effects occurred substantially early and faded at later intervals (Barr et al., 2013; Bates et al., 2015; Cai & Brysbaert, 2010; Dimigen, 2020; Hauk et al., 2006, 2009; Hauk & Pulvermüller, 2004; Hutzler et al., 2007; X. Li et al., 2011; Nasreddine et al., 2005; Sereno et al., 2020). Despite the growing body of research on Chinese reading, the neural dynamics underlying age-related differences in word length effects have yet to be examined using event-related potential methodology.

Several shortcomings of previous ERP studies should be highlighted. First, whether findings derived from isolated lexical decision paradigms adequately capture the complexity of contextual reading processes warrants further investigation. Second, the RSVP paradigm used in reading research interrupts natural attention deployment and parafoveal preview, and precludes word-skipping (Hutzler et al., 2007, 2013), thus diminishing ecological validity. To overcome these shortcomings, we simultaneously recorded EEG and eye movement data to examine age-related differences in word length effects during free-viewing Chinese reading. With this co-registration technique, EEG data are time-locked to the onset of first fixations on target words, thus generating fixation-related potentials (FRPs). The FRP waveforms resemble those of ERP and shed light on neural correlates of word processing in natural reading. To date, the application of co-registration methodology to investigate this phenomenon in ecologically valid Chinese reading contexts has not been reported.

This study employed EEG-eye movement co-registration to examine how aging modulates word length effects in Chinese text processing. The primary objective was to delineate the cognitive mechanisms responsible for the age-related increase in word length effects on fixation time measures. Specifically, this investigation examines whether aging influences the transition from word-based to character-based processing strategies among Chinese readers during the decoding of long words. We aimed to replicate larger word length effects for older adults compared to their younger counterparts on fixation time measures, with primary focus on the FRP data. This study addresses whether older adults adopt a character-based strategy for decoding long words, predicting reversed word length effects for older adults compared with young adults; otherwise, larger word length effects should be observed for older adults compared with young adults.

Methods

Participants

To ensure comparability with previous eye-tracking studies of Chinese sentence reading and word decision-making (S. Li et al., 2018; P. Liu et al., 2015), we employed a similar sample size to those studies. The data were collected between 2021 and 2022. A total of 46 participants were recruited from the local community, comprising 20 older adults (13 females, 7 males) and 26 younger adults (19 females, 7 males). All participants reported normal reading abilities and no history of cognitive, mental, or physical disorders. Older adults ranged in age from 60 to 73 years (M = 61.80, SD = 2.84), while younger adults ranged from 19 to 25 years (M = 20.04, SD = 1.51.

Visual acuity was screened using a Tumbling E chart for both groups, with no reliable group difference observed (older group: M = 4.99, SD = 0.09; young group: M = 4.97, SD = 0.11; t = 0.704, p > 0.05). No significant difference in years of education was found between older adults (M = 11.25, SD = 2.36) and younger adults (M = 12.12, SD = 1.97; t = 1.357, p > 0.05). All participants were required to be native speakers of Chinese and received ¥150 as compensation for their participation (approximately 21 USD). The Montreal Cognitive Assessment (MoCA) was administered to ensure normal cognitive functioning in older adults, with a minimum score of 26/30 required for inclusion (Nasreddine et al., 2005).

The Chinese-adapted Vocabulary subtest of the Wechsler Adult Intelligence Scale-Third Edition (WAIS-III) (Wechsler et al., 2003) was administered to assess vocabulary knowledge. An independent samples t-test revealed no significant difference between older adults (M = 14.85, SD = 0.93) and younger adults (M = 14.38, SD = 1.13; t = 1.487, p > 0.05). However, younger adults demonstrated significantly higher digit span scores (M = 14.35, SD = 1.52) than older adults (M = 11.55, SD = 1.73; t = 5.819, p < 0.001).

Design and Materials

The experiment followed a 2 (target word length: short vs. long) × 2 (group: younger vs. older adults) design. One hundred sixty-four framed sentences containing target words were used in this study. Examples of these sentences are presented in Figure 1 [FIGURE:1]. The complete set of materials can be accessed at the following Open Science Framework repository: https://osf.io/wxnk8/. Sentence stimuli varied in length from 12 to 22 characters (M = 17.45, SD = 1.93), and target words were systematically positioned at or near the midpoint of each sentence. Short condition target words comprised two characters, whereas long condition target words comprised four characters. Table 1 [TABLE:1] shows that short and long target words were systematically matched for character stroke frequency (wherein the third and fourth character positions in the short target word condition contained non-target characters; |ts| < 0.93, ps > 0.35). Moreover, word and character frequencies, calculated per million characters based on the SUBTLEX-CH corpus, demonstrated no significant differences between short and long target word conditions (|ts| < 1.02, ps > 0.31).

Twenty undergraduate participants rated sentence plausibility on a 5-point Likert scale, revealing no statistically significant difference between conditions involving long and short words (4.65 vs. 4.61, t = 1.227, p > 0.05). A cloze task involving 20 participants (10 younger adults and 10 older adults) was employed to assess target word predictability; none of these participants took part in the main experiment. The mean cloze probability for short words was 0.37% as rated by young adults and 0.24% by older adults. For long words, cloze frequencies were 0.43% as rated by young adults and 0.30% by older adults. A two-way ANOVA with word length (short vs. long) as a within-subjects factor and age group (younger vs. older adults) as a between-subjects factor revealed no reliable main effects of age group, target word length, or their interaction (Fs < 0.17, ps > 0.47).

A Latin square design was employed to counterbalance sentence frames, yielding two equivalent stimulus sets. Each set contained 164 sentence frames with an equal distribution of short and long target words. Participants were randomly assigned to stimulus sets within each age group. The experimental sentences in each stimulus set were administered in random order, with 12 practice sentences presented prior to the experimental trials. Comprehension was assessed using true/false questions administered after 5 sentences during the practice phase and after 40 sentences during the experimental phase. Age-appropriate response methods were employed: younger adults used keyboard input (right or left button presses) while older adults provided verbal responses.

Apparatus and Procedure

EEG signals were recorded using 32 electrodes connected to a BrainAmp amplifier (Brain Products, Germany) at a sampling rate of 1000 Hz. The AFz electrode served as the ground reference. Horizontal and vertical electrooculogram (EOG) channels were concurrently recorded to facilitate subsequent correction of ocular artifacts. Electrode impedances were maintained below 5 kΩ across all data collection sessions. Neural signals were digitized using an online band-pass filter with a frequency range of 0.1–100 Hz. Eye movement data were sampled at 1000 Hz using an EyeLink eye tracker system. Temporal synchronization between eye movement and EEG recordings was accomplished via transistor-transistor logic (TTL) pulses transmitted from the stimulus presentation computer at trial onset and offset. Fixation-related potentials (FRPs) were derived by time-locking EEG epochs to the onset of first fixations on target words. To ensure analytical consistency, identical fixation regions of interest were employed across both short and long target word conditions.

Testing was conducted individually. Upon arrival, participants provided written informed consent and completed a visual acuity examination to ensure normal or corrected-to-normal vision. The experimental protocol commenced with a practice phase consisting of ten trials to acclimate participants to procedural requirements. All stimuli were rendered in 20-point Song typeface and presented against a white background. Participants maintained a fixed viewing distance of 60 cm from the display monitor, ensuring that each character subtended approximately 1° of visual angle. Visual stimuli were displayed on a 19-inch LCD monitor configured at native resolution (1024×768 pixels) with a vertical refresh rate of 60 Hz. Prior to data collection, participants received instructions emphasizing reading for comprehension, followed by completion of a 3-point horizontal eye-tracking calibration procedure. Calibration accuracy was monitored before the initiation of each trial, with re-calibration implemented whenever observed accuracy exceeded the predetermined threshold of 0.5° of visual angle.

Data Analysis

Statistical analysis revealed that older adults exhibited significantly poorer comprehension accuracy than their younger counterparts (M = 84.0% vs. M = 96.5%; t = 5.215, p < 0.001). The analysis framework comprised two sets of dependent variables: first-pass processing measures (first fixation duration, gaze duration, target word-skipping probability, and refixation probability) and reprocessing measures (total reading time, regression path duration, probability of regressions-out, and probability of regressions-in). Following previous eye-tracking studies (S. Li et al., 2018), data preprocessing involved excluding fixations with extremely short and long durations prior to statistical analyses. Specifically, first fixation durations and gaze durations below 80 ms or exceeding 1200 ms were systematically excluded, affecting fewer than 3.9% of all experimental trials.

These measures were analyzed using linear mixed-effects models fitted with the lme4 package (version 1.1-30, Bates et al., 2015) in R (version 4.2.1). Linear mixed-effects models with maximal random structure were fitted to analyze eye-tracking dependent variables, incorporating full random-effects structures for both participants and items (Barr et al., 2013). When the maximal model failed to converge, model simplification was performed through systematic removal of random effects components, prioritizing item-level random effects before participant-level effects, with correlations among random effects removed first, followed by random slopes as needed. Word length, group, and their interaction were entered as fixed effects. Regression coefficients (b), standard errors (SE), t-statistics (t = b/SE), and p-values are reported below. To ensure robustness, parallel analyses were conducted using log-transformed continuous variables, which demonstrated identical patterns of statistical significance compared to non-transformed data. Given this consistency, only results from the non-transformed data analyses are presented.

EEG data were analyzed using the MATLAB toolbox EEGLAB (version 8.10). Raw EEG signals underwent offline preprocessing with a 0.1–40 Hz band-pass filter. To enhance detection of saccade-related artifacts, spike potentials were overweighted through replication of data within a temporal window extending from -20 to +10 ms around saccade events. A linked mastoid reference scheme (TP9/TP10) was applied to the raw EEG data during preprocessing. Ocular artifact correction was implemented using independent component analysis (ICA) methodology specifically adapted for fixation-related potential analyses (Dimigen, 2020). Independent components were rejected if their variance during saccadic epochs exceeded the corresponding fixation-epoch variance by more than 10%. Automated artifact rejection algorithms were applied prior to epoch averaging to exclude trials exhibiting excessive amplitude deviations (±100 μV) indicative of residual ocular contamination.

Event-related potentials time-locked to fixation onset (FRPs) were computed separately for each experimental condition and individual participant across epochs spanning -200 to +1000 ms relative to fixation onset. Mean amplitudes were quantified within theoretically motivated time windows: 100–300 ms for early components, 300–500 ms corresponding to the N400 response, and 500–800 ms encompassing the P600 component. Statistical evaluation of ERP data employed omnibus repeated-measures analyses of variance (ANOVAs) according to the analytical framework established by Sereno et al. (2020). The experimental design incorporated two within-subjects factors: Condition (2 levels) and Region of Interest (ROI). Electrode sites were systematically grouped into nine anatomically-defined ROIs corresponding to a topographical matrix (anterior-central-posterior × left-midline-right), as depicted in Figure 2 [FIGURE:2].

Results

Eye Movement Measures

Sentence-Level Analyses. Older adults spent more time comprehending sentences (4623 ms vs. 2966 ms), made more and longer fixations (fixation number: 15.9 vs. 11.4; mean fixation duration: 242 ms vs. 221 ms) and more leftward saccades (3.85 vs. 3.65), and made shorter forward saccades than their younger counterparts (2.21° vs. 2.95°). These age-related differences in sentence-level measures provide further support for previous research findings (S. Li et al., 2018; Z. Liu et al., 2017; Wang et al., 2018).

Word-Level Analyses. Descriptive statistics, including means and standard errors, along with statistical test results, are reported in Tables 2 and 3. Robust word-length effects were consistently observed across multiple dependent measures, with longer words (four characters) eliciting significantly extended gaze durations, regression path durations, and total reading times relative to shorter words (two characters). Furthermore, four-character words demonstrated elevated refixation probabilities and diminished skipping rates compared to two-character words, consistent with established word-length effects documented in the literature (S. Li et al., 2018; X. Li et al., 2011). Notably, first fixation duration exhibited a reversed word-length effect, with two-character words receiving longer initial fixations than four-character words, corroborating previous findings by Li et al. (2018). This phenomenon may reflect a cognitive processing strategy whereby both young and older adults exhibit reduced reliance on single prolonged fixations when decoding longer words compared to shorter words, as substantiated by the significant word-length effects on refixation probability.

Older adults exhibited systematically longer fixation durations across temporal eye-movement indices (first fixation duration, gaze duration, regression path duration, and total reading time) and demonstrated elevated refixation probabilities for target words relative to younger participants. Statistical analyses detected significant Group × Word Length interaction effects across eye-movement parameters. The interaction pattern was characterized by age-related differential responses to word-length manipulations: older adults exhibited amplified word-length effects on temporal measures (gaze duration, regression path duration, and total reading time) while showing attenuated effects on word-skipping probability compared to younger participants. Contrary to expectations based on prior research, the present study did not replicate the enhanced word-length effect on refixation probability previously documented in older adults (S. Li et al., 2018).

FRP Results

See Table 4 [TABLE:4] for statistical reporting and Figure 3 [FIGURE:3] for the mean amplitudes of word length differences in FRP amplitude for young and older adults.

ERP data in the 100–300 ms interval: The ANOVA demonstrated a significant main effect of ROI, reflecting a clear anterior-posterior voltage distribution with peak negative amplitudes over midline posterior scalp regions (-3.891 μV) and peak positive amplitudes over midline anterior scalp regions (2.771 μV). Further analysis of the Word Length × ROI interaction elucidated marginally reliable increased word-length effects on the amplitude of negativity over the midline posterior (-3.685 vs. -4.514 μV, p = 0.09), but numerically decreased word-length effects on negativity deflection over the left anterior (1.843 vs. 1.856 μV), left central (2.803 vs. 2.845 μV), midline central (1.107 vs. 1.178 μV), and left posterior (2.585 vs. 2.654 μV) regions. Post-hoc decomposition of the significant Group × ROI interaction revealed that younger adults demonstrated significantly more pronounced negative voltage responses over midline posterior scalp locations than older adults (-5.490 vs. -2.291 μV, p < 0.005), but statistically reliable decreased negativity over the left anterior (2.429 vs. 1.095 μV, p < 0.05), right anterior (2.354 vs. 1.135 μV, p = 0.06), midline central (1.742 vs. 0.363 μV, p < 0.05), left posterior (3.155 vs. 1.924 μV, p < 0.05), and right posterior (2.715 vs. 1.314 μV, p < 0.05) regions. Statistically reliable three-way interactions among group, word length, and ROI were observed. Further tests revealed an increased word-length effect on negative deflections over midline posterior for younger adults (-4.561 vs. -6.420 μV, p < 0.05), but decreased word-length effects across the nine scalp regions among the older group (ps < 0.05).

ERP data in the 300–500 ms interval: Statistical analysis yielded a significant main effect of ROI, demonstrating differential topographical voltage distributions: maximal negative deflections were recorded over midline posterior electrode locations (-4.472 μV), while maximal positive deflections were observed over midline anterior electrode locations (3.875 μV). Further testing of the Word Length × ROI interaction elucidated a reliable increased word-length effect on the amplitude of negativity over the midline posterior (-4.026 vs. -5.466 μV, p < 0.05), but no such reliable effects among other scalp regions (ps > 0.05). Further testing of the Group × ROI interaction revealed increased negativity over the midline posterior for younger adults compared to older adults (-6.573 vs. -2.372 μV, p < 0.005), but statistically reliable or numerically decreased negativity over the left anterior (3.312 vs. 1.601 μV, p < 0.05), midline anterior (4.105 vs. 2.858 μV), right anterior (3.057 vs. 1.440 μV, p < 0.05), midline central (1.877 vs. 0.216 μV, p < 0.05), right central (2.807 vs. 1.631 μV), left posterior (3.959 vs. 2.605 μV, p = 0.057), and right posterior (3.456 vs. 1.910 μV, p < 0.05) regions. Further tests clarified these three-way interactions, revealing that younger adults showed increased word-length effects on negative deflection over midline posterior scalp (-5.252 vs. -7.893 μV, p < 0.05), while older adults demonstrated decreased word-length effects over the midline anterior (2.625 vs. 3.091 μV, p = 0.051), right central (1.405 vs. 1.857 μV, p = 0.053), left posterior (2.269 vs. 2.941 μV, p = 0.043), and right posterior (1.602 vs. 2.218 μV, p = 0.068) regions.

ERP data in the 500–800 ms interval: We observed a reliable main effect of word length, with long words eliciting larger P600 responses than short words (-0.571 vs. 4.005 μV). A reliable main effect of ROI was also observed, with the smallest positive P600 over the midline posterior (-0.048 μV) and the largest P600 over the left anterior (3.212 μV). Further testing of the Word Length × ROI interaction elucidated a marginally reliable increased word-length effect on P600 amplitude over the midline posterior (-0.287 vs. 1.943 μV, p = 0.089), but reliable effects among other scalp regions (ps < 0.001). Follow-up contrasts examining the Group × ROI interaction revealed distinct age-related P600 distribution patterns. Younger adults demonstrated significantly attenuated P600 responses compared to older adults across lateral sites: left anterior (1.056 vs. 5.369 μV, p < 0.001), left posterior (0.234 vs. 0.995 μV, p < 0.05), and right central regions (1.327 vs. 3.750 μV, p < 0.05). Conversely, younger adults exhibited significantly enhanced P600 responses over midline sites: anterior (4.031 vs. 1.916 μV, p < 0.005) and posterior regions (1.280 vs. -1.375 μV, p < 0.001). No significant three-way interactions emerged for the P600 component.

In summary, we observed that younger adults showed relatively increased word-length effects on both negative deflections in earlier and N400 intervals over the midline posterior, whereas older adults demonstrated relatively decreased word-length effects on these negative deflections across the whole brain scalp, with long words eliciting smaller negative responses than short words. Additionally, long words elicited a larger P600 than short words, but three-way interactions were not persistent in the P600 component.

Discussion

In this study, we examined neural correlates of age differences in word-length effects by co-registering EEG data with oculomotor behavior during free-viewing Chinese reading. Our results regarding age differences in word-length effects align with previous eye-tracking studies and provide novel evidence from fixation-related brain potentials. Overall, we replicated older adults' larger word-length effects on viewing time measures such as gaze duration, regression path duration, and total reading time. These age differences in word-length effects can be interpreted as older adults preferring to use character-based strategies for Chinese comprehension (S. Li et al., 2018; Rayner et al., 2009) or requiring more effort to decode long words than their younger counterparts. Conversely, electrophysiological analyses failed to demonstrate enhanced word-length effects in the fixation-related potential data. Older adults displayed significantly reduced word-length sensitivity, indicating potential age-related shifts away from word-based lexical access strategies during the processing of extended orthographic strings.

Age-Related Changes in Early and N400 Word-Length Effects

Our FRP results for young adults were partially consistent with previous findings in alphabetic word processing (Loberg et al., 2019; Van Petten & Kutas, 1990). Specifically, we replicated reliable word-length effects from the earliest time window by observing increased negativity deflections over the midline posterior scalp for long words compared to short words, suggesting they process long words holistically. If larger word-length effects on fixation periods for older adults were due to greater effort in holistically decoding longer words, parallel larger word-length effects should appear in EEG data. As depicted in Table 4 and Figures 3 and 4, we found no evidence supporting older readers having larger word-length effects on either early negative deflection or the N400, thus completely refuting this account. Contrarily, decreased word-length effects on these brain responses were observed among older participants. Our results are not attributable to older adults' scaling properties of brain responses or their reduced brain activity (Wlotko et al., 2010). In summary, we revealed qualitative differences between older and young adults, supporting the view that older adult readers adopt different mechanisms from young adults when recognizing long words.

Our findings sharpen understanding of the language networks in older adults. Some researchers have argued that dynamic retrieval networks are susceptible to age-related decline, whereas lexical-semantic network structures are maintained throughout the adult lifespan (Jongman & Federmeier, 2022; Payne & Federmeier, 2017). In this study, word-length effects were used to tap into the language network particularities of older adults. Intriguingly, the effects of word length on earlier components over the posterior scalp suggested changes in the lexical-semantic network, whereas decreased effects over the anterior scalp may arise from age-related changes in dynamic retrieval networks. These findings resonate with perspectives that older adults' lexical-semantic networks are less organized, less connected, and less efficient (Krethlow et al., 2020). The interconnected lexical networks for long words are more sensitive to aging than those for short words, and/or during the N400 response, older adults have not yet assembled characters—especially for long words.

Our findings elucidate age-related shifts in cognitive processing mechanisms during reading comprehension. Converging evidence suggests that older adults demonstrate diminished capacity for context-driven, top-down processing strategies relative to younger adults, instead exhibiting increased reliance on stimulus-driven, bottom-up mechanisms for word identification (DeLong et al., 2012; Payne & Federmeier, 2017; Wlotko et al., 2010). Using visual half-field display procedures, Federmeier and Kutas (2019) established that older adults exhibited no left-hemisphere engagement in top-down, context-based processing, instead manifesting a compensatory transition to bottom-up processing mechanisms lateralized to the right hemisphere. Our results of decreased word-length effects over the right hemispheric scalp for older adults resonate with these views. Therefore, a possible neurocognitive reason for the lack of evidence supporting word-length effects on word identification among older adults is that they rely more on bottom-up interactive processing of characters for word recognition.

Overall, our early FRP results revealed that aging causes a shift from word-based to character-based strategies for decoding Chinese four-character words in reading. This may be due to older adults' visual declines (Owsley, 2011), inefficient parafoveal processing, and word segmentation difficulties (He et al., 2021; Zhang et al., 2019). Another possibility is that global representations in the mental lexicon for long four-character words are degraded, leading older adults to rely more on morpheme representations for recognition (Zhou & Marslen-Wilson, 2000). This account is supported by our FRP effects over posterior scalp regions. Moreover, brain aging may be the neural basis for this shift (Federmeier & Kutas, 2019; Jongman & Federmeier, 2022; Payne & Federmeier, 2017). Evidence supporting this comes from findings that decreased word-length effects were observed in both language dynamic retrieval and lexical-semantic networks, and that older adults relied on bottom-up right-hemisphere processing producing decreased word-length effects.

Non-Age-Related Changes in P600 Word-Length Effects

The present investigation builds upon existing research by strengthening the empirical foundation and advancing theoretical comprehension of compound word-length phenomena in Chinese reading processes. P600 effects are thought to be functionally associated with integrating semantic information into contextual representation (Aurnhammer et al., 2023; Delogu et al., 2019, 2021). In previous lexical decision studies of alphabetic languages, word-length effects dissipated and were even reversed in later intervals, with long words producing weaker P600 responses (Hauk et al., 2006, 2009). Inconsistent with these findings, we obtained increasing word-length effects over the midline posterior from early negative deflections to the P600 interval, suggesting more persistent word-length effects for Chinese word processing from pre-lexical processing to context integration. However, the absence of reliable three-way interactions on P600 suggests that older adults adopt a similar mechanism for integrating word meaning into contextual representation.

In summary, our study presents a comprehensive portrait of word-processing changes across aging groups. Young Chinese adults used a word-based strategy for processing both short and long multiple-character words; however, older readers adopted a character-based strategy for decoding four-character words while using similar mechanisms to young adults for integrating word meaning into contextual representation. We demonstrated that characters are an important decoding unit for Chinese people as they age, thus playing a significant role in word decoding. These results demonstrate a shift from word-based to character-based strategies for decoding long words as Chinese readers age. These findings align with established models of Chinese word recognition that incorporate both constituent character analysis and holistic word processing (Chen et al., 2003; X. Li et al., 2014; X. Li & Pollatsek, 2020; Yan et al., 2006), thereby elucidating the mechanisms by which Chinese readers decode compound words during text comprehension.

Limitations

Two limitations in our study should be acknowledged. First, we did not assess whether readers perceive four-character words as complete word units or treat them as two separate two-character words, though FRP data among young adults suggest they treat short and long target words as complete word units. The second limitation is that we co-registered EEG data to the first fixation on the first two characters of four-character words to examine how readers decode long words, thus missing data when the first fixation landed on the last two characters of four-character words. We did not separate temporal overlap components arising from different fixation points, nor did we sort out the EEG data for refixations. Future research should address these issues.

Declarations

Data Availability Statement

All data, computational code, and research materials have been made publicly available through the Open Science Framework platform (https://osf.io/wxnk8/). We acknowledge that this investigation was conducted without prior preregistration of the study design, hypotheses, or analytical procedures.

Ethics Approval and Consent to Participate

This investigation was conducted under the research project entitled "Study on the cognitive mechanisms of special population in Chinese reading." Ethical approval was obtained from the Institutional Review Board of the Cognition and Brain Disorders Research Centre at Hangzhou Normal University (Protocol No. 20190408), where all experimental procedures were conducted. The study was performed in accordance with the ethical standards outlined in the Declaration of Helsinki.

Consent for Publication

The data and ideas appearing in this manuscript have not been disseminated at any conference or meeting, posted on a listserv, or shared on a website. We do not borrow materials from other works. We grant exclusive rights to the publisher if the article is accepted for publication in the Journal of Research in Reading.

Competing Interests

There are no conflicts of interest to declare.

Funding Statement

This study was supported by Grants of Zhejiang Provincial Philosophy and Social Science Foundation (No. 23NDJC269YB).

Acknowledgements

Not applicable.

References

Aurnhammer, C., Delogu, F., Brouwer, H., & Crocker, M. W. (2023). The P600 as a continuous index of integration effort. Psychophysiology, 60(9). https://doi.org/10.1111/psyp.14302

Bai, X., Yan, G., Liversedge, S. P., Zang, C., & Rayner, K. (2008). Reading spaced and unspaced Chinese text: Evidence from eye movements. Journal of Experimental Psychology: Human Perception and Performance, 34(5), 1277–1287. https://doi.org/10.1037/0096-1523.34.5.1277

Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68(3), 255–278. https://doi.org/10.1016/j.jml.2012.11.001

Bates, D., Mächler, M., Bolker, B. M., & Walker, S. C. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1). https://doi.org/10.18637/jss.v067.i01

Cai, Q., & Brysbaert, M. (2010). SUBTLEX-CH: Chinese word and character frequencies based on film subtitles. PLoS ONE, 5(6). https://doi.org/10.1371/journal.pone.0010729

Chen, H., Song, H., Lau, W. Y., Wong, K. F. E., & Tang, S. L. (2003). Chinese reading and comprehension: A cognitive psychology perspective. In C. McBride-Chang & H. Chen (Eds.), Reading development in Chinese children (pp. 157–169). Westport, CT: Praeger.

Delogu, F., Brouwer, H., & Crocker, M. (2019). Event-related potentials index lexical retrieval (N400) and integration (P600) during language comprehension. Brain and Cognition, 135. https://doi.org/10.1016/j.bandc.2019.05.007

Delogu, F., Brouwer, H., & Crocker, M. W. (2021). When components collide: Spatiotemporal overlap of the N400 and P600 in language comprehension. Brain Research, 1766. https://doi.org/10.1016/j.brainres.2021.147514

Dimigen, O. (2020). Optimizing the ICA-based removal of ocular EEG artifacts from free viewing experiments. NeuroImage, 207. https://doi.org/10.1016/j.neuroimage.2019.116117

Federmeier, K. D., & Kutas, M. (2019). What's "left"? Hemispheric sensitivity to predictability and congruity during sentence reading by older adults. Neuropsychologia, 133. https://doi.org/10.1016/j.neuropsychologia.2019.107173

Gollan, T. H., Slattery, T. J., Goldenberg, D., Van Assche, E., Duyck, W., & Rayner, K. (2011). Frequency drives lexical access in reading but not in speaking: The frequency-lag hypothesis. Journal of Experimental Psychology: General, 140(2), 186–209. https://doi.org/10.1037/a0022256

Hauk, O., Davis, M. H., Ford, M., Pulvermüller, F., & Marslen-Wilson, W. D. (2006). The time course of visual word recognition as revealed by linear regression analysis of ERP data. NeuroImage, 30(4), 1383–1400. https://doi.org/10.1016/j.neuroimage.2005.11.048

Hauk, O., & Pulvermüller, F. (2004). Effects of word length and frequency on the human event-related potential. Clinical Neurophysiology, 115(5), 1090–1103. https://doi.org/10.1016/j.clinph.2003.12.020

Hauk, O., Pulvermüller, F., Ford, M., Marslen-Wilson, W. D., & Davis, M. H. (2009). Can I have a quick word? Early electrophysiological manifestations of psycholinguistic processes revealed by event-related regression analysis of the EEG. Biological Psychology, 80(1), 64–74. https://doi.org/10.1016/j.biopsycho.2008.04.015

He, L., Ma, W., Shen, F., Wang, Y., Wu, J., Warrington, K. L., Liversedge, S. P., & Paterson, K. B. (2021). Adult age differences in parafoveal preview effects during reading: Evidence from Chinese. Psychology and Aging, 36(7), 822–833. https://doi.org/10.1037/pag0000639

Hutzler, F., Braun, M., Võ, M. L.-H., Engl, V., Hofmann, M., Dambacher, M., Leder, H., & Jacobs, A. M. (2007). Welcome to the real world: Validating fixation-related brain potentials for ecologically valid settings. Brain Research, 1172(1), 124–129. https://doi.org/10.1016/j.brainres.2007.07.025

Hutzler, F., Fuchs, I., Gagl, B., Schuster, S., Richlan, F., Braun, M., & Hawelka, S. (2013). Parafoveal X-masks interfere with foveal word recognition: Evidence from fixation-related brain potentials. Frontiers in Systems Neuroscience, 7(July). https://doi.org/10.3389/fnsys.2013.00033

Jongman, S., & Federmeier, K. (2022). Age-related changes in the structure and dynamics of the semantic network. Language, Cognition and Neuroscience, 37(7), 805–819. https://doi.org/10.1080/23273798.2021.2019286

Kliegl, R., Grabner, E., Rolfs, M., & Engbert, R. (2004). Length, frequency, and predictability effects of words on eye movements in reading. European Journal of Cognitive Psychology, 16(1–2), 262–284. https://doi.org/10.1080/09541440340000213

Li, L., Bao, L., Li, Z., Li, S., Liu, J., Wang, P., Warrington, K. L., Gunn, S., & Paterson, K. B. (2024). Efficient word segmentation is preserved in older adult readers: Evidence from eye movements during Chinese reading. Psychology and Aging, 39(3), 215–230. https://doi.org/10.1037/pag0000813

Li, L., Li, S., Wang, J., McGowan, V. A., Liu, P., Jordan, T. R., & Paterson, K. B. (2017). Aging and the optimal viewing position effect in visual word recognition: Evidence from English. Psychology and Aging, 32(4), 367–376. https://doi.org/10.1037/pag0000163

Li, S., Li, L., Wang, J., McGowan, V. A., & Paterson, K. B. (2018). Effects of word length on eye guidance differ for young and older Chinese readers. Psychology and Aging, 33(4), 685–692. https://doi.org/10.1037/pag0000258

Li, X., Bicknell, K., Liu, P., Wei, W., & Rayner, K. (2014). Reading is fundamentally similar across disparate writing systems: A systematic characterization of how words and characters influence eye movements in Chinese reading. Journal of Experimental Psychology: General, 143(2), 895–913. https://doi.org/10.1037/a0033580

Li, X., Liu, P., & Rayner, K. (2011). Eye movement guidance in Chinese reading: Is there a preferred viewing location? Vision Research, 51(10), 1146–1156. https://doi.org/10.1016/j.visres.2011.03.004

Li, X., & Pollatsek, A. (2020). An integrated model of word processing and eye-movement control during Chinese reading. Psychological Review, 127(6), 1139–1162. https://doi.org/10.1037/rev0000248

Liu, P., Liu, D., Han, B., & Paterson, K. B. (2015). Aging and the optimal viewing position effect in Chinese. Frontiers in Psychology, 6(October). https://doi.org/10.3389/fpsyg.2015.01656

Liu, Z., Pan, Y., Tong, W., & Liu, N. (2017). Effects of adult aging on word encoding in reading Chinese: Evidence from disappearing text. PeerJ, 2017(1). https://doi.org/10.7717/peerj.2897

Loberg, O., Hautala, J., Hämäläinen, J. A., & Leppänen, P. H. T. (2019). Influence of reading skill and word length on fixation-related brain activity in school-aged children during natural reading. Vision Research, 165, 109–122. https://doi.org/10.1016/j.visres.2019.07.008

McGowan, V. A., White, S. J., Jordan, T. R., & Paterson, K. B. (2014). Aging and the use of interword spaces during reading: Evidence from eye movements. Psychonomic Bulletin & Review, 21(3). https://doi.org/10.3758/s13423-013-0527-8

McGowan, V. A., White, S. J., & Paterson, K. B. (2015). The effects of interword spacing on the eye movements of young and older readers. Journal of Cognitive Psychology, 27(5), 609–621. https://doi.org/10.1080/20445911.2014.988157

Nasreddine, Z. S., Phillips, N. A., Bédirian, V., Charbonneau, S., Whitehead, V., Collin, I., Cummings, J. L., & Chertkow, H. (2005). The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. Journal of the American Geriatrics Society, 53(4), 695–699. https://doi.org/10.1111/j.1532-5415.2005.53221.x

Owsley, C. (2011). Aging and vision. Vision Research, 51(13), 1610–1622. https://doi.org/10.1016/j.visres.2010.10.020

Payne, B. R., & Federmeier, K. D. (2017). Event-related brain potentials reveal age-related changes in parafoveal-foveal integration during sentence processing. Neuropsychologia. https://doi.org/10.1016/j.neuropsychologia.2017.10.002

Pollatsek, A., Rayner, K., Reichle, E. D., Stroud, M. J., & Williams, C. C. (2006). The effect of word frequency, word predictability, and font difficulty on the eye movements of young and older readers. Psychology and Aging, 21(3), 448–465.

Rayner, K., Castelhano, M. S., & Yang, J. (2009). Eye movements and the perceptual span in older and younger readers. Psychology and Aging, 24(3), 755–760. https://doi.org/10.1037/a0014300

Rayner, K., Slattery, T. J., Drieghe, D., & Liversedge, S. P. (2011). Eye movements and word skipping during reading: Effects of word length and predictability. Journal of Experimental Psychology: Human Perception and Performance, 37(2), 514–528. https://doi.org/10.1037/a0020990

Rayner, K., Yang, J., Schuett, S., & Slattery, T. J. (2013). Eye movements of older and younger readers when reading unspaced text. Experimental Psychology, 60(5), 354–361. https://doi.org/10.1027/1618-3169/a000207

Sereno, S. C., Hand, C. J., Shahid, A., Mackenzie, I. G., & Leuthold, H. (2020). Early EEG correlates of word frequency and contextual predictability in reading. Language, Cognition and Neuroscience, 35(5), 625–640. https://doi.org/10.1080/23273798.2019.1580753

Van Petten, C., & Kutas, M. (1990). Interactions between sentence context and word frequency in event-related brain potentials. Memory & Cognition, 18(4), 380–393. https://doi.org/10.3758/bf03197127

von der Malsburg, T., Kliegl, R., & Vasishth, S. (2015). Determinants of scanpath regularity in reading. Cognitive Science, 39(7), 1675–1703. https://doi.org/10.1111/cogs.12208

Wang, J., Li, L., Li, S., Xie, F., Liversedge, S. P., & Paterson, K. B. (2018). Effects of aging and text-stimulus quality on the word-frequency effect during Chinese reading. Psychology and Aging, 33(4). https://doi.org/10.1037/pag0000259

Wechsler, D., Chen, Y. H., & Chen, X. Y. (2002). WAIS-III Chinese version technical manual. San Antonio, TX: Psychological Corporation.

Wlotko, E. W., Lee, C., & Federmeier, K. D. (2010). Language of the aging brain: Event-related potential studies of comprehension in older adults. Linguistics and Language Compass, 4(8). https://doi.org/10.1111/j.1749-818X.2010.00224.x

Yan, G., Tian, H., Bai, X., & Rayner, K. (2006). The effect of word and character frequency on the eye movements of Chinese readers. British Journal of Psychology, 97(2), 259–268. https://doi.org/10.1348/000712605X70066

Zhang, J., Tong, W., & Liu, Z. (2019). The aging of perceptual span in different word length sentences: Evidences from eye movement. Psychological Development and Education, 35(3), 312–319. https://doi.org/10.16187/j.cnki.issn1001-4918.2019.03.08

Zhou, X., & Marslen-Wilson, W. (2000). Lexical representation of compound words: Cross-linguistic evidence. Psychologia: An International Journal of Psychology in the Orient, 43(1), 47–66.

Figure Captions

Figure 1. Examples of the target word manipulation in the experiment. Short target word condition: 整天幻想幸运光顾不如踏踏实实地工作 (English translation: "It is better to work hard than to fantasize yourself as a lucky person"). Long target word condition: 整天幻想幸运从天而降不如踏踏实实地工作 (English translation: "It is better to work hard than to fantasize yourself as a very lucky person"). Note: The underlined words are the targets; the target will not be underlined during reading.

Figure 2. Interest regions from crossing the dimensions of AntPost (Posterior, Central, and Anterior) and Hemisphere (Right, Midline, and Left).

Figure 3. Grand-averaged waveforms over nine scalp regions for young and older adults (negative voltage plotted upward).

Table Captions

Table 1. Means and standard deviations of word frequency, character frequency, and stroke number for target words.

Table 2. The mean and standard errors of eye-tracking measures in the target region.

Table 3. Statistical effects of the linear mixed-effects models for eye-tracking measures.

Table 4. F-results of ANOVA analyses as a function of age group, word length, and ROIs.