Empirical research on language aging remains inconclusive, with studies falling into two major categories of findings (Brysbaert et al., 2016; Cabeza, 2002; Goral et al., 2007; Kemper et al., 2003; Reuter-Lorenz and Cappell, 2008). A vast array of studies strokes the findings that language abilities were vulnerable to aging. Language decline would occur, though most often in differential degree, inevitably to elderly adults. For example, Rácz and Lukács (2023), by employing a visual lexical decision task, reported that elderly adults took longer to determine whether the visually presented word was familiar or not. Similar findings were obtained by Sabsevitz et al. (2005) and Thompson-Schill et al. (1997) although they respectively adopted a different lexicosemantic processing task, namely, semantic similarity judgment task and verb generation task. Rojas et al. (2024) reported an aging-related decline in accuracy in elderly adults' completion of lexical decision task. Specifically, they found that when elderly adults of different ages were asked to determine whether the presented word was real or not, those aged 80 years and above made more errors than those aged 60–69 years. Likewise, Miklashevsky et al. (2024) reported an aging effect in the elderly adults' processing of high vs. low motor-related action words. While the young adults responded faster and more accurately to high motor-related action words compared to low motor-related ones, this discrepancy was not found for elderly adults.

Another array of studies seemed to arrive at the finding that elderly adults do not necessarily undergo language degradation. They may, in some cases, even outperform younger adults. For instance, studies showed that elderly adults remained stable or even improved verbal knowledge (Bowles et al., 2005), and outperformed younger ones on vocabulary tests involving word definitions or multiple-choice formats (Verhaeghen, 2003; Bowles and Salthouse, 2008). Additionally, Rabaglia and Salthouse (2011) demonstrated that elderly adults' comprehension skills remained well preserved, especially when dealing with familiar or context-rich content. A common belief underlying these studies was that elderly adults accumulated more extensive knowledge over a lifetime, had greater semantic repertoire and lexical diversity, and underwent a wider experience with articulation and use of vocabularies (Dumas, 2015; Gollan and Goldrick, 2019). In other words, the lexicosemantic subsystem of language poses itself as an area that inherently combats against aging. The more people age, the richer semantic knowledge they develop (Barresi et al., 2000; Burke, 2006; Dumas, 2015; Gollan and Goldrick, 2019).

In sum, the contrasting findings in research on language aging underscores the uncertainty of this field, necessitating further inquiry into language aging.

To advance this line of research, the present study focuses on lexicosemantics—a core domain of language processing—and examines how aging may differentially affect the comprehension of action verbs based on the semantic granularity of their associated bodily effectors. According to embodied theories of language, understanding an action verb involves the simulation of sensorimotor content, and the explicitness with which a verb specifies its primary effector (e.g., eyes, mouth) may modulate the efficiency and automaticity of this simulation. This suggests that the effector specificity encoded in verb semantics could be a sensitive dimension to age-related cognitive changes across the lifespan.

To systematically investigate this possibility, we introduce a distinction between two classes of Chinese facial action verbs: overt and covert. This distinction is grounded in lexical semantics. Overt face action verbs (e.g., “睁眼” [to blink], “张嘴” [to open one's mouth]) explicitly encode a specific facial effector as an obligatory part of their core meaning, where the named body part is integral to the verb's lexical representation. In contrast, covert face action verbs (e.g., “观星” [to gaze at the stars], “吃面” [to eat noodles]) describe events directed by the face without lexically specifying a single, discrete facial effector. While such actions necessarily involve the eyes, the effector remains an implicit, backgrounded element within a broader event semantics centered on the target and experience.

Although in Chinese this semantic distinction often correlates with morphology—many overt verbs are compounds containing a body-part morpheme—the central theoretical claim pertains to the representational consequences of this difference in semantic explicitness. We hypothesize that overt verbs, by virtue of their lexically specified effectors, elicit more focused and automatic effector-specific simulation during comprehension (Pulvermüller, 2005; Shebani and Pulvermüller, 2018). Comprehending covert verbs, however, is hypothesized to require a more integrative and resource-dependent process to link the action to its implicit bodily substrate, which may result in slower and less efficient processing.

This theoretical framework leads to two primary, sequential objectives for the current study. First, we aim to establish the behavioral reality of the overt-covert distinction. We predict a main effect of Verb Type, with overt verbs being processed faster and more accurately than covert verbs in a deep semantic task, thereby validating this dimension as a significant moderator of lexicosemantic processing efficiency. Second, we use this novel distinction as a precise probe to investigate lexicosemantic processing in healthy aging. We derive two specific predictions: (1) Based on overwhelming evidence for general cognitive slowing, we predict a robust main effect of Age Group, with older adults showing longer response times (RTs) across the board. (2) More exploratorily, we examine whether the processing advantage for overt verbs is preserved, diminished, or enhanced with age. If aging disproportionately affects resource-demanding integrative processes, one might expect a larger overt-covert difference for older adults (an interaction). Conversely, if the semantic architecture supporting this distinction remains stable, the effect size may be age-invariant. By examining both main effects and their potential interaction, this study seeks to provide a nuanced answer to whether aging affects the lexical-semantic system in a generalized manner or in a way that is sensitive to the specific architectural features of verb meaning.

To this end, we employed a semantic categorization task that required younger and older adults to judge whether a verb denotes a body-executed action. This task forces meaning-based processing and is well-suited for testing our hypotheses. By comparing the performance of both age groups on overt and covert face action verbs, this study aims to advance our understanding of semantic aging, not merely by asking if it occurs, but by using a theoretically motivated distinction to explore how it manifests across different types of lexical-semantic representations.

A priori sample size estimation was performed using G^*Power 3.1 (Faul et al., 2007). The analysis showed that 17 participants would be needed to detect a large effect size (f = 0.40; Cohen, 1969) with 80% power for a two-way mixed ANOVA (α = 0.05, number of groups = 2, number of measurements = 4, non-sphericity correction = 1). A large effect size was specified because the primary aim of this study was to identify robust and theoretically meaningful age-related interaction effects, rather than subtle or marginal differences. This conservative approach was adopted to ensure sufficient statistical power to detect the effects of key theoretical interest while minimizing the risk of Type II errors. These calculations adhered to the guidelines for repeated measures ANOVA (a priori) provided in the G^*Power 3.1 manual (Faul et al., 2023) and Brysbaert (2019), suggesting a total sample size of 34, equating to 17 participants per group.

To account for possible exclusions, a total of 60 participants took part in this experiment, divided into two age groups: 30 young adults (mean age = 22.80 ± 3.95, range = 18–34; 11 males), and 30 elderly adults (mean age = 71.44 ± 4.38, range = 66–80; 11 males). The young adults were all students from Sichuan International Studies University (SISU), whereas the elderly adults were all retired teachers or ex-staff also from SISU. All participants were native Mandarin speakers and resided in mainland China. All of them were right-handed, with normal or corrected-to-normal vision, and none reported any neurodevelopmental, neurological, or psychiatric disorders. To assess the cognitive function of elderly participants and control for individual differences, all elderly participants were screened using the Montreal Cognitive Assessment (MoCA). Only those with MoCA scores ≥26 (indicating normal cognitive function) were included in the study. The elderly group had a mean MoCA score of 28.2 (SD = 1.5), with all participants scoring above the cutoff. This ensured that the elderly group as a whole was within the normal range of cognitive function, thereby minimizing potential confounding effects of cognitive decline-related individual differences on the experimental task. Written informed consent was obtained from all participants before the experiment, and they were compensated afterward.

Participants were administered a semantic categorization judgment task in Mandarin Chinese. The experiment adopted a 2 (Face Action Verb Type: overt face action verbs vs. covert face action verbs) × 2 (Age Group: young adults vs. elderly adults) mixed design.

The Face Action Verb Type factor included two levels. The first level was overt face action verbs, and the second level covert face action verbs. Both types of action verbs were composed of two Chinese characters, and they denoted or implied an action executed by face, specifically, either by eyes or by mouth. The difference lay in that one of the two characters included in the overt face action verbs denoted the effector which executed the verb action while covert face action verbs did not. For example, the overt face action verb, “合眼” [to close eyes], which comprised of the character “合” and the character “眼”, included the character “眼” [eye] which denoted the action-effector and provided a direct clue to the understanding of the eye-semantic component of the verb. In contrast, the covert face action verb, “观星” [to gaze at the stars], which comprised of the character “观” and the character “星”, did not include the character “眼” [eye] denoting the action effector and thus offered no clue to the understanding of the eye-semantic component of the verb.

Both of the face action verb types contained 20 items. These two types of face action verbs were matched in terms of character structure (both being “verb + noun”), number of strokes [t(19) = 0.889, p = 0.385] and mean frequency [t(19) = −0.924, p = 0.367].

For the task-setting purpose, 20 filler verbs were included, which were composed of two characters but denoted natural changes or happenings (e.g., “吹动” [to blow] and “风化” [to weather]), unrelated to human body actions. The filler verbs were selected based on the criterion that their primary and conventional meaning in modern, isolated Mandarin usage denotes a natural event or a non-corporeal action. While some filler verbs (e.g., “吹动”) could, in specific contexts, be associated with a human action, their dominant interpretation in the absence of disambiguating context is the natural one (e.g., “to be moved by the wind”). This selection strategy was employed to establish a clear and consistent “No” response category (i.e., actions not executed by a human body part) for the task, thereby reducing the potential for decision ambiguity during the experimental trials.

The Age Group factor also included two levels, i.e., the young adults and elderly adults.

Each participant completed a total of 60 trials, with an equal number of 20 trials for each type of verbs (i.e., 20 for overt face action verbs, 20 for covert face action verbs, and 20 for filler verbs).

Participants were seated in a comfortable chair approximately 50 cm from a computer screen. Each trial began with a fixation cross presented for 300 ms, followed by a blank screen for 300 ms. The target verb (either a face action verb or a filler verb) remained on the screen until the participant responded. Participants were instructed to determine as quickly as possible whether the target verb denoted an action executed by a human body part by pressing the “F” or “J” keys. The “F” and “J” keys were selected for their symmetrical placement, which supports a consistent hand posture and enables clean counterbalancing of response mappings. The key-to-response mapping was counterbalanced across participants, with half using “F” for “BODY-EXECUTED” and “J” for “NON-BODY-EXECUTED”, and the other half using the reverse configuration. Both accuracy and reaction times were recorded. The experiment was conducted in the Key Laboratory of Foreign Language Learning and Cognitive Neuroscience at SISU.

A two-way repeated-measures mixed-design ANOVA was performed on the ACCs. The main effect of Face Action Verb Type was significant [F (1, 58) = 7.622, p = 0.008, MSE = 0.966, η² = 0.116]. The ACCs of the overt face action verbs (98.33%) were significantly larger than those for covert face action verbs (95.67%). The main effect of Age group was also significant [F (1, 58) = 8.611, p = 0.000, MSE = 1.250, η² = 0.998]. The ACCs of the young adults (98.83%) were significantly larger than those of the elderly adults (95.17%). The interaction effect between Face Action Type and Age Group was not significant [F (1, 58) = 0.476, p = 0.493, MSE = 1.366, η² = 0.008]. These results indicated that the elderly adults tended to produce more errors in the semantic judgment task.

A two-way repeated-measures mixed-design ANOVA was performed on the RT data. The analysis revealed a significant main effect of Face Action Verb Type, F (1, 58) = 10.366, p = 0.002, MSE = 37.05, η² = 0.152. The RTs to the overt face action verbs (1,007.73 ms) were significantly faster than those to the covert face action verbs (1,127.03 ms; p = 0.000). The main effect of Age Group was also significant, F (1, 58) = 41.268, p = 0.000, MSE = 123.785, η² = 0.152. The RTs of the young adults (669.73 ms) were significantly faster than those of the elderly adults (1,464.98 ms; p = 0.000). The interaction between Face Action Verb Type and Age Group was not significant, F (1, 58) = 1.039, p = 0.257, MSE =52.042, η² = 0.022.

Given the non-significant interaction, the following comparisons are presented as exploratory analyses to illustrate data patterns and do not constitute formal statistical inference. These analyses showed that young adults responded faster than elderly adults for both overt (631.33 ms vs. 1,384.13 ms; p < 0.001) and covert (708.23 ms vs. 1,545.83 ms; p < 0.001) face action verbs. Within the young adult group, the mean RTs was lower for overt face action verbs than for covert ones (631.33 ms vs. 708.23 ms; p = 0.148). Within the elderly adult group, the mean RT was lower for overt face action verbs than for covert ones (1,384.13 ms vs. 1,545.83 ms; p = 0.003). The presence of the overt-covert effect found for the elderly adults, in contrast to the absence of this effect found for the young adults, might indicate an age effect as well. These results should be interpreted with caution. Table 1 displays the mean reaction times (ms) and accuracy rates (%) for each condition.

A further independent samples t-test showed that the RT discrepancy for the overt action verbs between the young adults and the elderly adults (1,444.97ms_{RT − elderly adults}−634.66ms_{RT − young adults} = 810.31 ms) was not significantly different from that (1,625.13ms_{RT − elderly adults} – 711.02ms_{RT − young adults} = 914.11 ms) for the covert action verbs [t (29) = −0.1.113 p = 0.275]. Both the overt face action verb and the covert face action verb exhibited an age effect, and the age effect for the former was equivalent to that for the latter.