How Easily a Word Sparks a Mental Image Shapes How Bilinguals Switch Languages

For the world’s more than half of people who speak two or more languages, every conversation involves a hidden act of mental juggling. Words from both languages are activated at once, and the brain must constantly select the right one while suppressing the wrong one. For decades, scientists have debated how tightly this bilingual language control is linked to general cognitive control — the same mental machinery that helps anyone ignore distractions and stay on task. A new study published in Current Psychology suggests that the answer depends on something surprisingly small: how easily a single word conjures a mental picture.

The research, led by Qingwei Xue of Peking University together with colleagues at Beijing Language and Culture University, the Hong Kong Polytechnic University, and the Max Planck Institute for Human Cognitive and Brain Sciences, examined fifty-six Chinese–English bilinguals. Participants completed a cued picture-naming language-switching task, in which they named pictures in either their first language (L1, Chinese) or second language (L2, English) depending on the color of the picture’s frame, and a Simon task, a classic measure of cognitive control in which people must ignore a stimulus’s location and respond only to its color. The team then used linear mixed-effects modeling to ask whether the properties of individual words changed both switching behavior and its tie to general cognition.

The property in question is lexical imageability — how readily a word evokes a mental image. Words like “sit” score high; words like “dream” score lower. According to Dual Coding Theory and the bilingual lexical-access framework, highly imageable words are anchored in rich, shared conceptual representations that connect smoothly across languages, so they can be retrieved through direct conceptual mediation. Less imageable words, by contrast, lean more heavily on language-specific lexical routes — in effect, the L2 word may be reached indirectly through its L1 translation. The researchers reasoned that this difference in processing pathway could fundamentally change how much domain-general cognitive control a bilingual needs to recruit when switching languages.

The behavioral results bore this out in striking detail. Across the board, switch trials were slower than non-switch trials — about 1,194 milliseconds versus 1,103 milliseconds — confirming the familiar switch cost. But the shape of that cost differed dramatically depending on imageability. For relatively higher-imageability verbs, switch costs were statistically symmetrical: roughly 83 milliseconds in Chinese and 99 milliseconds in English, with no significant difference between the two. For relatively lower-imageability verbs, the pattern flipped into a classic asymmetry, with switch costs of about 103 milliseconds in the dominant L1 versus only 70 milliseconds in L2 — the larger L1 penalty that the Inhibitory Control Model traditionally attributes to the extra effort of suppressing and then releasing a strongly activated dominant language.

That asymmetry matters because it has long served as a fingerprint of bilingual control. When the dominant language is harder to switch into than the weaker one — the so-called reversed language dominance effect — researchers typically infer that the dominant language required stronger inhibition. The new findings suggest this fingerprint is not a fixed property of a bilingual brain but depends on the words being produced. When the lexicon offers sturdy, image-based conceptual bridges between languages, control demands even out. When those bridges are weak, the dominant language’s lexical network reasserts itself, and control becomes lopsided.

The connection to general cognition proved equally conditional. The Simon score — the reaction-time penalty incurred when a square’s location conflicts with the required response — correlated positively with switch costs, but only for relatively higher-imageability words. There, the correlation was robust (r = .454, p = .001): participants with poorer Simon-task performance showed larger language-switching costs. For relatively lower-imageability words, the correlation essentially vanished (r = .071, p = .622). In other words, the same bilingual could look strongly coupled to domain-general control on one set of words and largely decoupled on another, within the very same task.

The authors interpret this through the lens of adaptive control. Under the adaptive control hypothesis, bilinguals flexibly recruit cognitive control depending on task and contextual demands. The new data extend that idea inward, to the level of individual lexical items. When lexical access runs through shared conceptual representations — as it tends to for highly imageable words — the operations involved overlap more with the stimulus–response conflict resolution measured by the Simon task, producing a visible covariance between the two control systems. When lexical access depends more on language-specific lexical pathways, switching draws on processes that the Simon task simply does not index, and the correlation dissolves.

The study’s design choices reveal how carefully such effects must be isolated. Because nouns and verbs differ systematically in how imageability behaves — verbs show greater variability in imageability ratings (SD = 0.32 versus 0.23 for nouns) — the team restricted materials to 32 bilingual verbs drawn from a standardized naming battery, median-split into relatively higher- and lower-imageability groups that were carefully matched on name agreement, visual complexity, and word frequency. Familiarity differences were statistically controlled. Participants’ third-language experience, switching habits, and L2 proficiency were likewise entered as covariates, and log-transformed reaction times were analyzed with maximal random-effects structures for both participants and items.

The authors are candid about the limits of their conclusions. Because the picture-naming paradigm required all stimuli to be nameable from images, the materials spanned only a restricted imageability range, and neither item-level regression nor trial-level sensitivity analyses found a continuous linear relationship between imageability ratings and switching behavior — the effect emerged only under the categorical contrast. Imageability norms were also drawn from a separate bilingual sample rather than the participants themselves, and the Simon task captures only one slice of cognitive control, specifically stimulus–response conflict rather than the monitoring, updating, and selection processes that language switching may also recruit. The externally cued paradigm, moreover, cannot speak directly to the voluntary code-switching that dominates real bilingual life.

Even with those caveats, the implications are considerable. Bilingualism research has often treated lexical materials as interchangeable, attributing variability in language control to proficiency, interactional context, or linguistic level. This study adds a fourth lever: the intrinsic imageability of the words themselves. If the coupling between bilingual language control and cognitive control can be strengthened or severed by nothing more than the mental imagery a word evokes, then decades of apparently conflicting findings may partly reflect unmeasured differences in stimulus properties. For theorists, it is a reminder that the bilingual mind is not one control system but a dynamic negotiation between domain-general machinery and the specific words it must wrangle — and that the picture in a word’s meaning can redraw the entire map of how two languages compete for the tongue.

Subject of Research: How lexical imageability modulates bilingual language switching and its relationship with cognitive control

Article Title: Relative lexical imageability modulates the relationship between bilingual language control and cognitive control

Article References: Xue, Q., Zhang, Z., Liu, Y., Zhao, Y., & Chen, L. (2026). Relative lexical imageability modulates the relationship between bilingual language control and cognitive control. Current Psychology, 45(17), Article 1444. https://doi.org/10.1007/s12144-026-09908-7

Image Credits: AI Generated

DOI: 10.1007/s12144-026-09908-7

Keywords: bilingualism, language switching, cognitive control, lexical imageability, psycholinguistics, switch costs, Simon task, dual coding theory, lexical access, Chinese-English bilinguals, inhibitory control, Current Psychology