<p>Understanding the cognitive mechanisms of spoken language production is essential for explaining human communication. Grounded in embodied cognition theory, this study investigates how object-directed sensorimotor features shape the conceptual preparation stage underlying concept selection. In Experiment 1, participants selected concepts from two dimensions: graspability (graspable vs. non-graspable) and semantic category (artificial vs. natural). This was done under three levels of response competition. Results showed a robust processing advantage for graspable objects over non-graspable ones across all tasks. More importantly, a systematic and pervasive advantage for artificial objects over natural ones was also found specifically in the control and low-choice task conditions. Experiment 2, which also employed the concept selection paradigm, examined graspable objects only, distinguishing between precision and power grips. Results revealed that precision-grip objects were selected faster than power-grip objects, an advantage that was most pronounced in the low-choice condition. Notably, this precision motor facilitation was reliable for both artificial and natural objects. Moreover, the advantage for artificial objects over natural objects remained significant and was independent of the grip-type effect. These findings demonstrate that motor representations are automatically and robustly activated during conceptual preparation. Their facilitatory influence is characterized by a consistent categorical advantage for artifacts and an additive benefit from fine-grained motor parameters, with the expression of these effects being dynamically calibrated, but not abolished, by task-imposed cognitive demands. This study underscores the hierarchical and context-sensitive interplay between sensorimotor and semantic systems in concept selection during spoken language production.</p>