Conceptualization can be defined as “the process of abstracting the real world into the concepts we use to refer to what is there” (Williams et al., 2012, p. 51). Concepts are the “glue” that connects our past, present, and future (Murphy, 2002). They can be considered cognitive and mental representations of categories we form to comprehend, navigate, and interact with the physical and social environment.

Understanding how people conceptualize the natural world and use geographical and geopolitical categories is becoming increasingly urgent considering the impact on our daily lives of environmental transformations stemming from the climate change emergency and international political contrasts. Unfortunately, although several studies have explored the impact of the ecological crisis on human wellness, focusing, for instance, on emotional (eco-anxiety - Clayton, 2020; Pihkala, 2020) or physical eco-consequences (e.g., Abraham et al., 2023) and identified factors leading pro-environmental behaviors (Li et al., 2019) and strategies to incentivize them (Schultz, 2014; Tam et al., 2021), only a few studies focus on how people represent ecology-related phenomena (e.g., Baquiano and Mendez, 2015, 2016; Falcinelli et al., 2023; Mbaye et al., 2023; Dézma et al., 2024). In addition, scarce attention is dedicated to investigating concepts referring to places in which climate change events typically occur—i.e., geographical kinds such as mountains, oceans, forests, and geopolitical entities such as state, city, metropolis. In our opinion, this is a critical issue to address, considering that human choices and behaviors are influenced not only by emotional, personological, and contextual aspects, but also by people’s perception and representation of reality (Myers, 2007).

To that end, our review focuses on geographical and geopolitical concepts—from now on, “geo” concepts—with the aim of discussing and summarizing the most influential studies carried out on the topic. We start by defining the domain of inquiry and clarifying the method we used to select the papers. We then detail the rationale behind the choice of focusing on geo concepts, and we outline possible shortcomings that might have motivated the scarcity of studies in this area. We then touch upon cross-cultural, cross-linguistic, and interindividual variability from a theoretical and methodological perspective, looking at how this might shape geo concepts. Most studies, indeed, have emphasized the variability of these concepts when compared to other conceptual domains (e.g., Battig and Montague, 1969; van Overschelde et al., 2004; van Putten et al., 2020) and across different languages and cultures—both industrialized and rural—and current positions tend to favor a non-universalistic characterization of them (e.g., Burenhult and Levinson, 2008; Mark et al., 2011; Turk and Stea, 2014). In addition, the cognitive demand requested by specific tasks (e.g., Mark et al., 1999, 2001; Smith and Mark, 2001a,b; Pires, 2005), idiosyncratic characteristics of individuals such as their level of expertise (e.g., Giannakopoulou et al., 2013; Wartmann et al., 2014; Purves et al., 2023), as well as the typology of the inhabited environment (e.g., Williams et al., 2012; Wartmann et al., 2015) further impact geo conceptualization. Thus, geo concepts well-exemplify the flexibility of the human conceptual system (e.g., Barsalou, 1993; Connell and Lynott, 2014; Mazzuca et al., 2021, 2022). The last section of the review discusses how information about the places we live in is stored and organized in our semantic memory (e.g., Tversky and Hemenway, 1983, 1984; Patton, 1995; Lloyd et al., 1996), leveraging studies inspired by the classical theory of hierarchical levels (Rosch and Mervis, 1975; Rosch et al., 1976, 1978).

Importantly, all the reviewed studies have accessed geo-concepts through the words that express them. Results from the reviewed studies are summarized in Table 1.

This review concerns concepts related to the natural environment. While this is a broad conceptual domain, we narrowed our analysis to two classes of concepts following specific selection criteria. We excluded studies focused on the conceptualization of individual natural kinds like animals (e.g., tiger, bee, cat) and plants (e.g., olive tree, flower, oak) since these domains have already been extensively investigated (e.g., ethnobiology and folk biology studies - see Medin and Atran, 2004; Atran and Medin, 2008). Literature on ecological concepts—i.e., concepts referring to the ecological emergency (e.g., climate change, ozone hole, deforestation)—is still limited (for a few exceptions, see Baquiano and Mendez, 2015, 2016; Falcinelli et al., 2023; Mbaye et al., 2023; Dézma et al., 2024), preventing us from including this class of concepts in the review.

Our review focused instead on geographical and geopolitical concepts, i.e., concepts indicating locations where people usually live and maintain social relations. Specifically, geographical concepts (e.g., mountains, forests, oceans) refer to large-scale physical entities and places in which human intervention has been minimal over time, thus remaining well-preserved in their original forms (i.e., the “totally natural” categories from Mausner, 1996; see also Pasca et al., 2018). Conversely, geopolitical concepts (e.g., cities, states, metropolis) include geographical elements highly affected by human intervention, such as urbanization, and politically relevant entities (i.e., the “non-natural” categories from Mausner, 1996; see also Pasca et al., 2018).

In our review, we will use the term “geo” to refer jointly to both classes, as in most of the studies we targeted, these were treated together. Instead, we will use specific labels (i.e., geographical and geopolitical) when illustrating studies explicitly focusing on one of the two domains. Notably, the definition we just provided for environmental places—based on the criteria of the degree of human intervention in transforming the natural environment—and their distinction into geographical and geopolitical concepts pertains to a specific discipline and research field, i.e., psychology and psycholinguistics (Mausner, 1996; Pasca et al., 2018). In other fields, such as geography, but also political science, and sociology, the investigation of geographical and geopolitical concepts happens jointly, and their distinction does not exist or is not so marked. The divergence in definitions across disciplines is, per se, informative since it further testifies to the variable character of the geo domains, which does not only concern the linguistic and cultural differences, as we extensively shown in the course of the review, but also extends to scientific disciplines.

To retrieve relevant articles and materials on the conceptualization of geo entities, we conducted an exhaustive literature search between December 2021 and April 2022, with a subsequent bibliographic search integration in July-August 2023 and April 2024 to include more recently published materials.

A first bibliographic search was performed on Google Scholar and PubMed, with a second double-check made on Scopus. In our searches, we used the conjunction of words that refer to geographical and geopolitical entities (e.g., “nature”, “geographical”, “geopolitical”), with labels denoting their specific status (e.g., “concepts”, “categories”, “features”), and/or with terms indicating tasks typically used in categorization research (e.g., “free-listing”, “ethnophysiographic methods”), or research field typically dealing with categorization (“ethnophysiography”, “ontology”, “hierarchical organization”), and/or with words referred to conceptual variability (e.g., “variation”, “flexibility”, “malleability”) and/or with words related to individual characteristics, such as expertise levels (e.g., “folk”, “commonsense”, “expertise”). Specifically, we used the combination of the following terms: (“green” OR “geographical/geography” OR “geopolitical/geopolitics” OR “natural/nature” OR “natural environmental/environment” OR “ecological/ecology”) AND (“categories” OR “concepts” OR “terms” OR “features” OR “categorization” OR “conceptualization” OR “linguistic encoding” OR “features encoding”, “entities”) AND/OR (“linguistic task” OR “semantic fluency (task)” OR “feature listing (task)” OR “free-listing (task)” OR “property generation (task)” OR “feature generation (task)” OR “category norms” OR “rating (task)”, OR “natural language analysis” OR “ethnophysiographic method(s)”) AND/OR (“ethnophysiography” OR “ontology” OR “ontological investigation” OR “Geographical Information System/GIS”, “hierarchical organization” OR “hierarchical levels”) AND/OR (“variation” OR “variability” OR “flexibility” OR “conceptual flexibility” OR “linguistic relativity” OR “malleability”) AND/OR (“native theory” OR “folk” OR “commonsense” OR “expertise” OR “proficiency” OR “non-expert” OR “laypeople”).

All terms were searched as keywords within the text and as words belonging to the title and/or abstract. We did not apply any restriction on the publication date range—due to the scarcity of contributions on the topic—and we only considered publications in English.

To increase the likelihood of including as many relevant studies as possible, we conducted a further manual search. We inspected the list of references of the publications we retrieved and searched for relevant materials on the most influential authors’ Google Scholar or Scopus Profile.

The literature search yielded a total of 116 papers. Materials consisted of papers in peer-reviewed journals, chapters of books, and unpublished Ph.D dissertations.

To evaluate the relevance and eligibility of articles, we used a hierarchical approach. The total list of papers was first assessed for duplicates. Then, we screened documents based on their title and abstract and excluded those that did not fit our research topic. The remaining articles were examined more in-depth by reading the entire manuscript, and those that met the inclusion criteria were included in the review. We selected articles that: 1) focused on the conceptual representation of geo concepts—i.e., geographical and geopolitical concepts—rather than on that of other natural entities, like plants and animals, or ecological issues; 2) included studies with designs and results based on human-participants testing rather than computational models; 3) involved linguistic tasks (e.g., ratings, free-listings, ethnophysiographic methods), to avoid potential confounding factors derived by heterogeneous tasks’ requests. Based on such criteria, we excluded 66 documents, thus resulting in a total of 50 articles eligible for the present review. Figure 1 shows a flowchart of the steps we implemented to select the final set of papers we will present in the review.

Understanding how people conceptualize geographical entities, such as mountains, seas, forests, and geopolitical ones, such as states, regions, and boundaries, is crucial for several reasons.

Before illustrating them, it is important to mention some research lines that characterize literature on concepts and categorization and that are relevant to the present review. Among these, some studies have focused on general dimensions that might differentiate concepts, like their degree of concreteness∼abstractness (e.g., Villani et al., 2019), i.e., to what extent concepts evoke the five senses and refer to specific, spatially bounded objects (Borghi and Binkofski, 2014). On top of that, some researchers started to study the domains for which this distinction might be problematic—because it might shift depending on the context (e.g., gender: Mazzuca et al., 2020, 2023; olfaction: Majid et al., 2018). A further important line of research is represented by studies assessing conceptual content and focusing on properties of specific conceptual domains. For example, much behavioral, neuropsychological, and neuroscientific research has been dedicated to investigating differences in the processing and brain representation between living and non-living entities (for reviews: Warrington and Shallice, 1984; Humphreys and Forde, 2001; Forde and Humphreys, 2005), or artifacts and natural kinds (e.g., Keil, 1992), and more recently on concepts that are a sort of hybrid between artifacts and natural objects, i.e., food concepts (Rumiati and Foroni, 2016; Papies et al., 2020; Mazzuca and Majid, 2023). These research strands are worth mentioning as they are the ground for substantiating the claim that geo concepts might be considered hybrid concepts, too (see section “6.3 Two paradigmatic cases of geographical variability: the concepts of Landscape and Forest”).

What are the reasons that make studying geo concepts particularly important?

First, it might contribute to understanding how people act on and interact with the natural and social environments. Specifically, it could illuminate how people live inside environmental transformations caused by climate changes and political contrasts.

Second, investigating geo concepts is paramount because of their specific cognitive profile. As we will show during the review, these concepts refer to large-scale physical elements. Still, because geographical and geopolitical features are not pre-segmented by nature, the role of categories imposed by humans is pivotal to defining them. Consequently, the geo conceptualization might be more variable than that of other semantic domains (Mark et al., 2010; van Putten et al., 2020), thus representing an excellent example of conceptual flexibility (e.g., Barsalou, 1993; Connell and Lynott, 2014; Mazzuca et al., 2021, 2022).

Two further reasons make the study of geo concepts particularly important. From a scientific standpoint, examining concepts either grounded in physical reality or determined by cultural and linguistic differences could facilitate the construction of an ontology of the geographic and geopolitical domains. In addition, research in this area also has a practical slant. Defining an ontology of the geo domains is fundamental for supporting non-specialist use of Geographic Information Systems (GIS), like Google Earth, that assist human activities such as navigation, resource management, and emergency services (Kuhn, 2001). GIS are usually accessed by heterogeneous groups of people with different sociocultural and linguistic backgrounds, interests, and age-related specificities. Hence, addressing the question of the universality of geographic and geopolitical concepts is crucial when dealing with GIScience (Mark and Turk, 2003; Burenhult and Levinson, 2008; Turk and Stea, 2014). To maximize their performance, GIS should account for the substantial variability in encoding geographical and geopolitical features in different languages and cultures. For instance, Google Earth maps names of places worldwide, risking forcing indigenous names to a universal ontology derived from an English conceptualization of landscapes (see also Blasi et al., 2022). Locals may carve up reality in quite different ways, evidencing the limits to the efficient use of these tools (Mark and Turk, 2003; Burenhult and Levinson, 2008; Turk and Stea, 2014). In the framework of contrasting the climate change crisis, this understanding can provide a conceptual basis on which to identify and implement more idiosyncratic policies aimed to take actions to contrast climate change consequences and enforce policies for sustainable development, leveraging on how people of different cultures conceive the environment surrounding them. In addition, it can contribute to facilitating the revitalization of global partnerships for sustainable development, considering cultural and linguistic differences (e.g., see Goals 13, 14, 15, and 17 of the United Nations 2030 Agenda for Sustainable Development - United Nations, 2015).

Despite their importance, gaining knowledge about geographical and geopolitical concepts is quite demanding. In particular, the intrinsic characteristics of these domains make their study particularly challenging.

First, people conceptualize geographical and geopolitical entities ambiguously, both in terms of objects (“what” concepts) and places (“where” concepts - Burenhult and Levinson, 2008). This aspect has been thoroughly examined by scholars who analyzed the encoding of “what” and “where” concepts in different aboriginal languages, such as Makalero (a Papuan language of East Timor - Huber, 2014, 2018), Marquesan (an Austronesian language spoken on the Marquesas Islands - Cablitz, 2008), and Lokono (an Arawakan language of the Guianas - Rybka, 2015, 2016). These studies revealed that “what” nouns typically denote individuated, discrete (Lyons, 1977, p. 693), relatively small and potentially moveable entities (Talmy, 1972, 2000, p. 315), such as Persons, Animals and Physical Objects (Lyons, 1977) that are typical exemplars of the superordinate “Thing” category (e.g., chicken, blanket, stone, or John - Huber, 2018, p. 478; Mackenzie, 1992, p. 256). Conversely, “where” concepts commonly denote entities that can be defined as large and stationary, such as Artificial and Natural Locations that are common exemplars of the superordinate “Place” category (e.g., office, church, mountain, cliffs - Huber, 2018, p. 478–480; Stolz et al., 2014, p.42). While in most cases, the referents of “what” nouns possess clear physical boundaries (Lyons, 1977; Jackendoff, 1983; Langacker, 2013), the referents of “where” nouns mostly lack clear perceptual boundaries that are readily identifiable (Cablitz, 2008; Rybka, 2015, 2016).

In this framework, an interesting perspective on geo concepts came from Lyons (1977, p. 693), according to whom artificial and natural locations might grant a special status of indeterminacy between the “what” and “where” categories, allowing for the possibility of being treated as one or the other in different languages. Take, for example, the concept of forest. In the sentence “The forest is huge”, the forest is perceived as an object located on the surface of the Earth. On the contrary, in the sentence “The ruin is in the forest”, the forest becomes a place, i.e., a part of the Earth’s surface where we move (Lyons, 1977, pp. 477). Likewise, the concept of the city can be conceived both as an object we talk about (e.g., “This city is called Rome”) and as a location (e.g., “About 40 thousand people live in that city”). Therefore, the same geo concept can belong to different ontological categories depending on the meaning arbitrarily assigned to it (Jackendoff, 1983; Landau and Jackendoff, 1993), hence displaying different conceptual properties.

To integrate this evidence, Stolz et al. (2014) showed that “Places” objects (e.g., office, school, church, Stolz et al., 2014, p. 42) were the only nouns that did not need an overt locative marker (“zero-marking” - Huber, 2018) to be identified as places—thereby representing intrinsically “where” nouns, i.e., nominal predicates with an intrinsic locative interpretation. Moreover, it was found that zero-marking of natural location nouns was a typical pattern across languages—it was present in several investigated aboriginal languages (Stolz et al., 2014; Huber, 2018). In contrast, “what” concepts (i.e., those referring to “Thing” objects such as chicken, blanket, stone, or John - Mackenzie, 1992, p. 256) can be used only with semantically specific locative verbs to denote locations since they do not have an intrinsic “where” interpretation.

To summarize, concepts like mountains, lakes, and cliffs assume a double ontological status—both of object and places—while this is not the case of other conceptual domains.

The lack of clear perceptual boundaries of natural and artificial locations has a further implication related to the criteria we use to determine entity boundaries, making studying these domains particularly hard. Geographical entities such as valleys, mountains, or lakes are parts of the Earth’s surface, which is an objectively continuous surface (Smith and Mark, 2001a, p. 93). Differently from plants and animals, landforms more properly belong to continua (Mark et al., 2010). Due to this, landscape features are not, for the most part, pre-segmented by nature but by human categorization processes. In other words, the lack of clear perceptual boundaries for natural entities makes human decisions rather than perception the most salient criterion to determine their status. For example, we can think of a valley in different ways, e.g., “as a concave fold between mountain ranges, or as the flat bottom of such a fold, or as the entire drainage area right up the flanks or do so without any such concept” (Burenhult and Levinson, 2008, pp. 137–138). Similarly, boundaries of artificial entities such as states, cities, and nations—despite being often physical like walls, water bodies, and fences—do not pre-exist; they are usually defined by political conventions. A direct consequence is that geographical and geopolitical concepts might be more variable across individuals and more affected by contextual factors such as language, culture, environment, and individual perspectives than other domains (Mark et al., 2010). In line with this, for instance, research has shown that different languages segment and outline landforms differently (e.g., Burenhult and Levinson, 2008; see section 6.3 “Two paradigmatic cases of geographical variability: the concepts of Landscape and Forest”).

Finally, the strong interconnection and mutual influence between the geographical and geopolitical domains makes it hard to disentangle them sharply and thus study them separately. Take, for example, national borders. Political forces usually define them, but natural features can also contribute by providing physical limits. For instance, the water surrounding the land could induce governments to decide that “Italy” ends where the sea starts. Likewise, political interests can transform a geographical element like a valley into a geopolitical entity such as a city or, similarly, a coast into a seaport. While the physical referent is the same, the concept endorses different ontological statuses depending on human interventions.

To sum up, the ambiguous nature of geographical and geopolitical concepts, their high variability, and their peculiar relationship with each other and their referents make their study particularly complex. We argue that these aspects should be considered to make cross-cultural communication more efficient and enhance governmental policies aimed at increasing sustainability in different countries.

Whether speaking different languages differentially shapes human cognition is still debated in cognitive science (Lucy, 1997; Casasanto, 2008; Whorf, 2012; Majid et al., 2018; Blasi et al., 2022). Relativist accounts contrast with nativist ones, according to which cognitive processes remain invariant regardless of specific languages and cultures. While several studies showed that language has a pervasive influence on cognition (Malt and Majid, 2013; Goddard and Wierzbicka, 2014; Kemmerer, 2019, 2022), such an influence might be more marked for some domains. It has been proposed elsewhere (Borghi and Binkofski, 2014; Borghi et al., 2018a,b, 2019a,b; Borghi, 2019, 2023) and presenting some data showing (Borghi and Mazzuca, 2023; Da Rold et al., 2023) that linguistic variability might be more pronounced for abstract than for concrete concepts. In the context of the debate between relativists and universalists, geo concepts might represent an interesting case. Indeed, several studies have demonstrated that their conceptualization—and in particular that of geographical entities such as landscape, water bodies, and other geographical features—varies more across languages and cultures than other conceptual domains (such as animals or body parts - see van Putten et al., 2020). Researchers have addressed this issue using a multitude of approaches. Two methodologies have dominated this literature so far, namely free-listing tasks and methods from ethnophysiography. Free-listing tasks are semantic fluency tasks that collect the frequency and type of properties generated for specific categories of concepts (e.g., Battig and Montague, 1969; van Overschelde et al., 2004; van Putten et al., 2020). Instead, ethnophysiographic methods analyze the encoding of geographical features in natural languages (e.g., Mark et al., 2010; Turk et al., 2012).

In the research field we examined, free-listing was employed to study the categorization of geographical and geopolitical entities, while ethnophysiographic methods primarily focused on geographical ones.

In the following subsections, we will first separately focus on these two methodologies and the geo-flexibility results they yielded (sections “6.1 The variability of geo concepts: evidence from free-listing tasks” and “6.2 The variability of geo concepts: evidence from ethnophysiographic methods”). We will then report evidence on conceptual variability for two geo concepts, i.e., landscape and forest (section 6.3 “Two paradigmatic cases of geographical variability: the concepts of Landscape and Forest”). Finally, we will discuss the reasons that underlie the emergence, variable in nature, of geo concepts (section 7).

While cross-linguistic evidence suggests geo conceptualizations consistently vary across sociocultural contexts, most languages seem to possess terms that denote geographical features—although partitioning the environment in culture-specific ways. This raises the question about the origins of these mental categories. Burenhult and Levinson (2008) identified three main forces: I) the salience of geographical features, from both a perceptual and cognitive perspective; II) utilitaristic affordances, i.e., the benefits that a particular place brings to the life of a community; and III) cultural and linguistic models of a group (see also studies in ethnobiology for similar proposals: Hunn, 1982; Berlin, 1992; Malt, 1995). The weight of these factors could diverge across domains (Huber, 2014; Mazzitelli, 2018).

Through the analysis of Indigenous languages, several studies in ethnophysiography unveiled the importance of these three aspects as milestones for categorizing (e.g., Brown, 2008; Burenhult, 2008; Mark et al., 2010, Turk et al., 2012; Huber, 2014; Mazzitelli, 2018). For instance, Mazzitelli (2018) compared English with Nalik, an Austronesian language spoken in the New Ireland Province (Papua New Guinea), and found that Nalik’s categorization of landscape features was tightly linked to their affordances (i.e., action opportunities - Gibson, 1977): for instance, daanim “freshwater” (and by extension “river”, “creek”) and raas “salty water” (and by extension “sea”) were not distinguished by size or shape, but exclusively by their chemical properties, and hence by the use the community can make of them (one can drink daanim but not raas). Along with affordances, perceptual salience—i.e., the visual prominence of landscape features that makes them easily recognizable and perceived as distinct elements (as an elevation as opposed to flat terrain)—also played an important role. For example, the distinction between raas (“shallow sea on the reef”) and laman (“open sea”) can be traced back to their visual properties (the open sea is deeper and darker than the shallow sea on the reef), as well as to their affordances (in the open sea different fishes are found than on the reef). Finally, the cultural meaning of places and their importance to the community’s life were a further categorization criterion, driving the attribution of proper names to geographical features. For instance, river sources, canoe passages, caves, and villages all had a proper name because, according to Nalik’s beliefs, they were where spirits lived. Instead, even perceptually salient entities like rivers and smaller hills mostly lacked proper names and were referred to using the name of the land they flowed through or were located in Mazzitelli (2018).

Additionally, the likelihood of a feature being easily communicated can further underlie landscape categorization. Regier et al. (2016) proposed that different ways to categorize the landscape are the effect of adapting languages to the physical environments where they are spoken to facilitate efficient communication among native people. They tested this hypothesis by analyzing the encoding of snow/ice terms in different speech communities. It has been suggested that Eskimo/Inuit languages possess more terms to indicate different subtypes of snow (e.g., aput “snow on the ground”, qana “falling snow”) than other languages such as English (Martin, 1986; Whorf, 2012), hence partitioning more sharply the notion of snow compared to other languages—although this idea has been largely questioned by other evidence (e.g., Pullum, 1989, 1991). Regier et al. (2016) empirically tested this hypothesis by analyzing multiple sources of data, such as reference works, Twitter, and large digital collections of linguistic and meteorological archives. They aimed to understand whether geographic position and temperature could predict the presence of different terms for snow and ice in a given language. Specifically, they predicted a link between warm temperatures and a unique semantic category encompassing ice and snow. Results aligned with these predictions: languages with separate terms for ice and snow were spoken in cold and warm regions. By contrast, languages that collapse this distinction in their lexicons were spoken exclusively in warm regions. This asymmetry is consistent with the view that the need for efficient communication probabilistically shapes language. In line with this, there should be less pressure to preserve the distinction—and thus a stronger tendency to collapse the two concepts—in warm than cold countries. Notably, this phenomenon characterizes not only Eskimo/Inuit languages but also other languages spoken in cold climates, such as Russian (Krupnik and Müller-Wille, 2010).

Finally, the specific structural properties of languages can also influence how people think about landscapes. O’Meara and Bohnemeyer (2008) investigated this aspect by comparing English and Seri (Mexico) terms for geographical features. The authors found that Seri’s lexicon was more complex than English and characterized by a prevalence of descriptive-analytical terms. Seri people linguistically defined land and water forms primarily according to their material composition and other features: shape, orientation, spatial, and physical features. O’Meara and Bohnemeyer (2008) proposed that this complex “model” for forming geographical terms might be due to the peculiarity of the Seri language rather than to specific ways of conceptualizing geographic entities. Indeed, complex expressions similar in structure to those for landscape features were widespread in Seri and extended to natural kinds and artifact terms. This, however, does not rule out the possibility that the linguistic structure of Seri landscape terms might have cognitive consequences. Indeed, the analytic linguistic system of Seri might lead native speakers to pay more attention to the material and spatial properties of landscape entities compared to speakers of languages with monomorphemic geographical terms, such as English.

Overall, all these studies suggest that the conceptualization of the geographic domain is not universal, and differences can emerge because of the interplay of different factors. Some point to the importance of perceptual or cognitive salience of geographical features, their action opportunities (i.e., affordances), and human cultural beliefs and models (Burenhult and Levinson, 2008). Others instead emphasize the centrality of communication efficacy in specific environmental settings (Regier et al., 2016). In addition, specific structural characteristics of a language are an additional criterion to consider when drivers of categorization are investigated (e.g., O’Meara and Bohnemeyer, 2008).

While in previous sections we tackled cross-cultural and cross-linguistic divergences in geo conceptualization, here we focus on other possible sources of variability. Specifically, we show how individual levels of expertise, the task’s cognitive demand, the experimental setting, and its familiarity can differentially affect geo conceptualizations.

This final section deals with the information people store about the places they inhabit and how this content is organized in semantic memory, i.e., a kind of human long-term deposit that conserves information about the world (Tulving, 2011; Reilly et al., 2023). We will illustrate studies inspired by the classical theory of hierarchical levels (Rosch and Mervis, 1975; Rosch et al., 1976, 1978). According to this proposal, semantic information is organized in a three-level conceptual hierarchy. Superordinate concepts (e.g., animal) are maximally distinctive but not very informative; subordinate concepts (e.g., bulldog) are very informative but not very distinctive. Finally, intermediate, basic level concepts (e.g., dog) would be the most frequently used, as they maximize informativeness and distinctiveness, as well as represent those containing a greater amount of stored information.

Patton (1995) extended Rosh’s view to the geo domain by proposing a three-level hierarchy in which geo-information is stored at a subordinate, basic, or superordinate level. The nodes in the subordinate level correspond to actual places a person has experienced, and the basic-level nodes represent more abstract categories of places. For example, when we visit a location like Rome, we learn about that specific city (subordinate level) and the basic level category city, which is more abstract. At the highest hierarchical level (i.e., superordinate) is the most abstract spatial concept, i.e., the category of place.

Along these lines, Lloyd et al. (1996) were interested in assessing the amount of information deposited at the superordinate, basic, and subordinate levels for geo categories. They asked participants to list as many characteristics, activities, or parts they associated with a particular geo term as possible. Characteristics were defined as “attributes that could be used to describe something”; activities were defined as things that “reflect responses or behaviors”; parts were defined as “things you can see and interact with” (Lloyd et al., 1996, pp. 187–188). The sample was divided into 11 groups: 5 groups were asked to generate terms (characteristics, activities, or parts) for the subordinate level categories: “home county”, “home region”, “home state”, “home city”, and “home neighborhood”. Other 5 groups were asked to generate terms for the basic-level categories: “country”, “region”, “state”, “city”, and “neighborhood”. Finally, a third group was asked to generate terms for the superordinate category “place”. The sample consisted of 630 students from introductory geography classes at the University of South Carolina, Columbia, and the University of North Carolina, Greensboro. In line with Rosch’s view (Rosch and Mervis, 1975; Rosch et al., 1976, 1978), results showed a substantial increase of information between the superordinate and basic levels and relatively little difference between the basic and subordinate levels for all three types of information. Therefore, the main argument regarding the amount of information stored at each hierarchy level from Rosh’s theory seems to be strongly supported by geo categories.

Tversky and Hemenway (1983) also conducted several experiments on this topic. In such experiments, they used “indoors” and “outdoors” as superordinate categories. Basic level categories were “home”, “school”, “store”, and “restaurant” for the indoor scene and “park”, “city”, “beach”, and “mountains” for the outdoor scene. The subordinate categories were specific cases like “elementary school” for “school” or “Midwestern city” for “city”. The first set of experiments provided 210 participants with both photographs and verbal descriptions of environmental scenes at three levels of abstraction and had them give lists of attributes, activities, and parts appropriate for each scene. Results indicated that scenes representing superordinate categories (“indoor” and “outdoor”) shared very few attributes, activities, and parts, suggesting that the scenes were very distinctive. Basic-level scenes (e.g., “beach”, “mountains”, “homes”, “school”) were less distinctive than superordinate categories, i.e., they shared many attributes, activities, and parts, but also more informative, as participants could list significantly more features. Scenes for more specific categories at the subordinate level (“elementary school” and “lake beach”) did not share significantly more properties than scenes representing basic-level categories—suggesting they were only slightly less distinctive than basic-level scenes. Because the number of activities, attributes, and parts listed for subordinate scenes only slightly exceeded that for the basic-level scenes, they were also somewhat more informative than basic-level scenes. Among the various types of items listed by participants, Tversky and Hemenway (1984) argued that parts proliferated in participants’ listings for basic-level categories. Still, few parts were listed for superordinate categories. This aligns with previous literature on hierarchical levels of categorization (Rosch et al., 1976, 1978; Hemenway, 1981; Murphy and Smith, 1982). In this regard, Tversky and Hemenway suggested that members of basic-level categories can be distinguished by their parts. Still, members of subordinate categories share parts and differ in other attributes. In the second set of studies (Tversky and Hemenway, 1983, 1984), the authors asked participants to provide labels for photographs of scenes and to complete sentences such as “The Kingstons furnished their ______ with furniture they built themselves”. Results revealed that participants preferred basic-level terms. Indeed, they used more frequently basic level terms when labeling photographs of scenes and when completing sentences describing activities performed in scenes, even though more specific or more general terms would have been appropriate too. Thus, the conceptual hierarchical level preferred in communication corresponds to the level determined to be basic under perceived appearance and perceived behaviors. In conclusion, the authors showed that knowledge about scenes and objects has a hierarchical organization in which an intermediate level of abstraction is preferred in cognition, behavior, and communication measures.

Overall, findings confirm that, like other semantic domains, geo entities are stored according to the hierarchical levels of categorization proposed by Rosch and colleagues (Rosch and Mervis, 1975; Rosch et al., 1976, 1978). They, therefore, highlight the necessity to consider the difference between concepts’ hierarchical levels in efficient governmental communication on sustainability, referred to the geo domain.

For decades, the conceptualization of geo—i.e., geographical and geopolitical—entities has been at the center of debate among scientists—especially those in the geographical and psychological fields—which is still ongoing. At the core of the discussion, there is the question of whether the categorization can be considered universal or if linguistic and cultural differences influence it. While some approaches defended the idea of homogeneity of geo classification (e.g., Council of Europe, 2000), most of the positions illustrated in this review strongly support the thesis of relativity (non-universality) of these categories. Indeed, numerous pieces of evidence showed that geo categories consistently vary across contexts, cultures, and languages (e.g., Burenhult and Levinson, 2008; Turk and Stea, 2014) and are even more variable than other semantic domains (e.g., Battig and Montague, 1969; van Overschelde et al., 2004; van Putten et al., 2020). Importantly, geo-meaning is also influenced by many other potential factors, such as people’s level of expertise (e.g., Giannakopoulou et al., 2013; Purves et al., 2023; but see also Wartmann et al., 2014), the kind of implemented task (e.g., Mark et al., 1999, 2001; Smith and Mark, 2001a,b; Pires, 2005), living environments along with their perceived familiarity (e.g., Williams et al., 2012; Wartmann et al., 2015). So, our review indicates that geo concepts are a highly flexible class of words.

This evidence suggests that geo concepts might possess a hybrid character. Indeed, although they usually refer to concrete entities—i.e., referents that can experience through actions and senses (e.g., Brysbaert et al., 2014)—as typical concrete concepts, their high variability makes them similar also to more abstract words (for linguistic/cultural variability of abstract concepts, see e.g., Borghi et al., 2018a,b, 2019a,b; Borghi, 2019, 2022; Borghi and Mazzuca, 2023; for variability in expertise, see e.g., Croijmans et al., 2020; Villani et al., 2022).

The malleability of this special domain can be partially explained by the fact that geographical and geopolitical features are not pre-segmented by nature but depend on how we look at them to categorize their exemplars. This is a direct consequence of their intrinsic nature: their referents, like valleys and coasts, or human transformations of them, like cities or ports, are not merely located in space but are also typically parts of the Earth’s surface, which is a continuous surface.

Investigating geographical and geopolitical concepts might add further and essential evidence to laypeople’s overall perception of the natural environment. This might be helpful for the definition of a domain ontology and inform the improvement of geographical information systems, like Google Earth, that assist in everyday human activities, such as navigation. It might also contribute to raising consciousness about how people experience the environmental damages caused by recent political contrasts and the climate change emergency (Baquiano and Mendez, 2015, 2016; Malt and Majid, 2023) and, vice versa, how pre-existing representation of places they inhabit might affect the (lack of) ecological and political actions for contrasting them.

Crucially, assessing whether and to which extent geo meaning varies according to cultural and linguistic aspects can provide a conceptual basis for identifying and implementing more idiosyncratic governmental policies on these topics that can allow communicating more effectively and thus reach as many people as possible. For instance, promoting effective policies for sustainable development (as those, for instance, described in the United Nations 2030 Agenda - United Nations, 2015) requires effective communication, which, in turn, involves a deep comprehension of how people of different countries conceptualize their world and can be motivated to change it.

The knowledge of the various ways laypeople conceptualize the geographical and geopolitical domains and the recognition that languages, cultures, people’s expertise, and the environments where they live deeply influence their conceptualizing represents, in our view, a critical piece allowing the promotion of a better world.

IF: Conceptualization, Resources, Writing – original draft, Writing – review and editing. CF: Writing – review and editing. CM: Writing – review and editing. AB: Conceptualization, Funding acquisition, Writing – review and editing.