Do Language Cues Shape Spatial–Numerical Associations? A Test of SNARC and MARC in Chinese–English Bilinguals

Spatial-numerical associations (SNAs) are often attributed to language, yet cross-cultural findings are confounded by reading direction and broader experience. We tested whether the language-of-presentation modulates the SNARC (small-left/large-right) and MARC (odd-left/even-right) effects using a within-participant Chinese-English bilingual design with number words. In Experiment 1 (magnitude classification; N=60 adults), the SNARC effect emerged in both Chinese (L1) and English (L2); slope-based analyses indicated comparable strength across languages, while a trial-level linear mixed model (LMM) detected a stronger effect in L1. MARC was not reliably observed in either language using both slope-based analysis and trial-level LMM. In Experiment 2 (parity judgement; N=52 adults), SNARC was not reliably observed in either language using both slope-based analysis and trial-level LMM. The slope-based MARC effect did not reach significance , but a trial-level LMM revealed a significant MARC effect in both languages that was stronger in L1. Individual-difference analyses revealed associations between language experience, cognitive variables, and spatial-numerical effects, with inhibitory control moderating the language difference in MARC.. These results place a lower bound on language-specific modulation of spatial-numerical mappings when format and reading direction are controlled: language-of-presentation influences these mappings, with effects stronger in L1 for both SNARC and MARC. Notably, these effects were detectable through trial-level modelling but not traditional slope-based analyses, highlighting the importance of analytical approach in spatial-numerical research. Individual differences in language experience and cognitive abilities may additionally shape the strength of these effects.