Abstract
Recent evidence has indicated that spatial representations, such as large-scale geographical maps, can be retrieved from natural language alone through cognitively plausible distributional-semantic models, which capture word meanings through contextual relationship (i.e., non-spatial associative-learning mechanisms) in large linguistic corpora. Here, we demonstrate that spatial information can be extracted from purely linguistic data even at the medium-scale level (e.g., landmarks within a city). Our results indeed show that different spatial representations (i.e., with information encoded either in terms of relative spatial distances or absolute locations defined by coordinate axes) of the underground maps of five European cities can be retrieved from natural language. Furthermore, by selectively focusing on the London tube, we show that linguistic data align effectively with both geographical and schematic visual maps. These findings contribute to a growing body of research that challenges the traditional view of cognitive maps as primarily relying on specialized spatial computations and highlight the importance of non-spatial associative-learning mechanisms within the linguistic environment in the setting of spatial representations.