Do all people, from all cultures and all languages, think about geographic space and geographic processes in more or less the same way? Or are there significant cross-cultural variations in how different peoples conceptualize and reason about geographic processes, features and places? Do geography and spatial relations parallel the infamous case of the many "Eskimo [Inuit] words for snow," the linguistics factoid that every cocktail party conversationalist thinks he or she knows? (The snow words situation turns out to be mostly a misinterpretation, but that's another story!) Or is spatial cognition and related linguistic development governed by universal principles?

Coming closer to home, we can ask similar questions about GIS. Are GIS software products and Spatial Data Infrastructures (SDIs) the same for all, a universal foundation based only on the true nature of geographic phenomena, universal principles of computing and cognitive primitives? Or are the GIS packages and SDIs that we know today biased toward a European worldview and a so-called 'Western' scientific approach? If GIS is universally easy-to-use (or universally difficult, but GIS usability is another topic!), then that is good news both for humanitarians and for software vendors, and we can move forward with a one-size-fits-all solution to the world's geospatial problems. But if GIS is biased toward the culture that produced it, then it could be yet another case of North Atlantic Imperialism, another brick in the darker side of globalization, posing an ethical dilemma, especially for those working in indigenous GIS or GIS and international development.

Mark Twain once wrote: "There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact." We have very few facts about cultural differences in spatial cognition. For almost 20 years, I have been intrigued by questions of possible cultural biases in GIS software, but until recently I had not found much in the way of solid answers. I now think I had been looking for cultural differences in spatial cognition in all the wrong places, in at least two different dimensions. I'll explain, and then draw some implications for GIS.

In the 1990s, I wrote numerous papers on spatial cognition, GIScience and GIS. A frequent justification for spatial cognition research was that it relates to GIS usability and especially to GIS universality. Papers at Latinamericanist and Latin American GIS meetings speculated that cultural and linguistic differences between Spanish and English might lead to added barriers to easy GIS use by Spanish speakers. But testing human subjects on spatial relations and language in the United States, Spain and Costa Rica failed to reveal significant differences for line-region spatial relations. Was the concern unfounded?

On sabbatical in 2002, I was fortunate indeed to have the opportunity to visit Andrew Turk at Murdoch University in Perth, Australia. Turk, in turn, took me a thousand miles north to Roebourne, Australia, where about 500 Yindjibarndi speakers were keeping their language alive after being forcibly removed from their traditional lands almost a century ago. Through semi-structured interviews and other ethnographic methods, Turk and I learned details of the Yindjibarndi terms for features of the landscape, things that in English would be called hills and valleys, pools and cliffs, gullies and riverbeds.

While a dictionary published in the 1980s appeared to show term-by-term translations, we found that in most (if not all) cases, the terms and their definitions did not line up! There was a many-to-many relation between Yindjibarndi terms for elevated areas and their English equivalents. The relation between rivers and their beds was turned inside out in this tropical desert area. Rivers in English are fundamentally composed of water, even if some are sometimes dry; whereas a wundu in Yindjibarndi is a (dry) channel that on rare occasions might contain water. When water is present after the rare heavy rains following a tropical cyclone, the water is categorized by its intensity of flow and is always distinct from the permanent feature that might get named "river" in English. The conceptual systems that underlie the semantics of geographic expressions in the Yindjibarndi language do not seem strange, but they definitely are different. Working with Turk, David Stea, Carmelita Topaha, and many Navajo friends, we have found similar differences in Navajo language landscape categories in the arid highlands of northern Arizona and New Mexico.

These tendencies do not appear to be unique to the languages we have examined to date. Discussions with researchers at the Max Planck Institute for Psycholinguistics (Nijmegen, The Netherlands) indicate that they are finding similar results for a wide variety of languages, cultures and environments: terminology for landforms, as well as systems of place naming, vary considerably between peoples living in different sorts of environments, in different cultural traditions and lifestyles.

Interesting perhaps, but how does this relate to GIS? Spatial data infrastructures encode entity types or feature codes in order to enhance the semantics of geospatial data. But such codings might not add value from an indigenous perspective, unless data are also encoded according to the categorical systems of the indigenous people. We might try to avoid categories altogether and just store entities and attributes, and then try to infer categories later on. But theories from cognitive science tell us that inferring categories from observable properties will be difficult, if not impossible, to implement.

All of the above might raise ivory tower curiosity and fuel esoteric dissertation research. But these issues have important theoretical implications for indigenous GIS and indigenous mapping. If there are significant cultural differences in any aspect of spatial cognition or language, surely this must in turn imply that current GIS is Eurocentric. As it turns out, unmodified commercial GIS has proved to be a valuable and at times powerful tool for indigenous people. Just as maps were powerful tools of European imperialism in the 18th and 19th Centuries, GIS returned the favor by supporting aboriginal land claims with presentations in a language that dominant-culture courts could appreciate. And many tribes in the western United States use unmodified commercial GIS to manage tribal lands and maximize productivity of forests and grazing lands, just as private land holders or government agencies might do. But in most (if not all) of these, the GIS was wielded by tribal members with university degrees or by consultants from the dominant culture. It is not clear whether tribal GIS experts can think in their own language and culture while using GIS, or whether traditional and spiritual values can be incorporated directly into GIS solutions.

Earlier, we found few differences in spatial conceptualization for spatial relations, but more recently we believe we have solid evidence that spatial feature categories can be quite different across languages. We were also working mainly comparing English with Spanish and French, other Indo-European languages. Moving to geographic entity types rather than spatial relations, to non-Indo-European languages in Australia and the western United States, and to arid landscapes rather than humid, we still see broad cross-language similarities but also many differences in the details.

Indigenous GIS faces all of the challenges of "Geographic Information and Society," a research agenda within Geographic Information Science. But indigenous GIS also faces additional challenges due to the cognitive and linguistic factors mentioned above. GIS for community empowerment requires qualitative methods and ease of use, whether in Indian country, in low-income urban settings or in suburbia. Socioeconomic and educational barriers to GIS use and the steep "Digital Divide" may be greater than the cultural and linguistic issues raised here. But if there is any truth in our recent findings, indigenous GIS will never be on an equal footing with, or fully interoperable with, the main stream of GIS until indigenous spatial cognition and conceptualization are understood as well in the indigenous cultures in question as they are for English and the dominant culture of North America and Europe.