The Chinese Room argument describes a thought experiment that suggests that for symbols to become meaningful, they must be grounded in perceptual experiences. Embodied cognition theorists frequently use this argument to claim that cognition requires perceptual simulation. We shed light on the symbol grounding problem by arguing that the structure of natural language provides language users with cues that allow them to bootstrap meaning from non-grounded symbolic co-occurrences, such that the statistical linguistic structure can bootstrap meaning with minimal grounding. Two studies show that co-occurrences of both Chinese and Arabic city names can reliably predict their longitude and latitude in China and in the Middle East. Using the statistical linguistic technique Latent Semantic Analysis, similarity ratings were obtained for Chinese city names (Study 1) and for Arabic city names (Study 2). Multidimensional scaling (MDS) coordinates of these similarity ratings correlated with the actual longitude and latitude of these cities, showing that cities that are located together share similar semantic contexts. These results suggest that the Chinese Room argument might be substituted with a Chinese Route argument: statistical linguistic frequencies of word co-occurrences provide language users with implicit cues about how to form perceptual representations.