terça-feira, 27 de junho de 2017

The ecology of spatial relations: the case of Kriol prepositions

Hildo Honório do Couto

Universidade de Brasília

1. Introduction

This paper seeks to examine the prepositions in the Portuguese Creole of Guinea-Bissau (henceforth, Kriol) in the framework of the relationships that obtain between language and environment. With a view to studying this type of relationship, some European linguists have proposed, in the last decade of the 20th century, what has come to be called ecolinguistics (Fill 1993), although the idea in itself goes back to as early as the Greek philosophers. For instance, in the famous dialogue by Plato, Cratylus, Cratylus claims that there is an influence of the things of the world in the words that designate them (physis). Hermogenes defends the opposite view, i.e., that words are formed by convention (thesis). In 1912, Edward Sapir recognized the influence of the environment in the lexicon, although he was reluctant to see the same kind of influence in grammar (in Mandelbaum 1949: 89-103). In 1970, Einar Haugen also dealt with the subject language-environment, in an essay dedicated to what he called "ecology of language" and "language ecology" (see Haugen 1972).   

With respect to "language and environment", most researchers emphasize only one type of language environment, namely, the social environment, as is the case with Haugen (1972) as well as with Fill (1993). Other kinds of environment are mentioned only in passing. In fact, we can detect at least four types of language environment. The first one is what I have been calling the Fundamental Environment of Language, better known as Fundamental Ecosystem/Ecology of Language. The other three emerge from inside it. They are the Social, the Mental and the Natural Environment of Language. Given that they are a prerequisite, especially the Natural Environment, to the study of the ecology of spatial relations as applied to prepositions, I give a relatively detailed presentation of them.

In section 2, I present the new field of study ecolinguistics, the science of the relationships between language and environment. In section 3, I present the Fundamental Environment of Language, which unfolds in the Social (3.1), the Mental (3.2) and the Natural (3.3) Environment of Language. Subsection 3.4 discusses some of the possible interrelationships that obtain among the previous three language environments. Section 4 is dedicated to the Ecology of Spatial Relations in general, illustrated with Portuguese spatial prepositions. In section 5, I present an overview of Kriol prepositions, as a preamble to the discussion of spatial prepositions. Section 6 focuses upon the ecology of spatial prepositions in Kriol, with reference to which prepositions fill the spatial relations shown in section 4. In section 7, I investigate other spatial prepositions not included in section 4, namely, movement prepositions. In section 8, I deal with prepositions that are apparently not spatial, although, in the end, they, too, are reducible to spatiality. Section 9 deals with the question of whether there are temporal prepositions, unconnected to spatial prepositions. Section 10 looks at what Bernard Pottier calls the notional use of prepositions. Section 11 discusses some uses of spatial prepositions that sound strange to speakers of European languages. In section 12, I discuss the question of whether the so-called "complex prepositions" are really so or are dividable into smaller constituents. In section 13, I try to relate the spatial relations examined in the previous sections with the physical environment of language, since it is generally thought that prepositions as grammatical words do not have anything to do with the world outside language. In Section 14, some final remarks are given.

2. Ecolinguistics

The term ecolinguistics came up only in the eighties, but the domain it studies dates back at least to 1911, when Edward Sapir gave a talk under the title of "Language and environment" at the American Anthropological Association. In the following year it appeared as an article in American Anthropologist 14, p. 226-242. Later on it was reprinted in Mandelbaum (1949). But, one of the first conscious uses of "ecolinguistics" to designate a specific discipline was made by Fill (1993) and, independently, by Makkai (1993). The former defines it as follows: "Ökolinguistik ist jener Zweig der Sprachwissenschaft, der den Aspekt der Wechelwirkung berücksichtigt, sei es zwischen einzelnen Sprachen, zwischen Sprechern und Sprechergruppen, oder zwischen Sprache und Welt, und der im Interesse einer Vielfalt der Erscheinungen und Beziehungen für die Bewahrung des Kleinen eintritt" (Fill 1993: 4). Considered the father of ecolinguistics, Einar Haugen, had previously defined what he called "ecology of language" and "language ecology" as "the study of the interactions between any given language and its environment" (Haugen 1972: 325).

Haugen's previous definition is not incompatible with Fill's, since the latter includes in his definition the idea that "ecolinguistics is [...] the branch of linguistics that [...] investigates the interrelationships [...] between language and world (emphasis added)". Following these leads, I define ecolinguistics as the discipline that studies the relationships between language and environment. This definition takes Haugen's literally. However, we could ask: What is a language environment? Given that ecolinguistics is epistemologically rooted in ecology, it is important to start with ecology's basic concept, namely, the ecosystem (Odum 1971). The nearest equivalent I was able to find is the relationship between language, the people that use it and the territory where these people live. In the last years, I have been calling this ecosystem Fundamental Environment/ecosystem of Language, as discussed in the following section. When we focus upon this ecosystem, we see that it comprises three sub-ecosystems inside it. The first and most conspicuous is the Social Environment of Language. The second is the Mental Environment. The third, the Natural Environment. Let us examine each of them in turn.  

3. The Fundamental Environment of Language

When a lay person hears about an unknown language (L), what first comes to his/her mind is the question: Which people (P) speak it? This people, of course, must live in a specific territory (T). In other words, common sense leads us directly to what has come to be called the Fundamental Ecosystem/Ecology of Language, or Fundamental Environment of Language (FEL), since the ecosystem is nothing more than the environment and the organisms that live in it, in their inter-relationships, taken together. These terminological variants do not have any bearing on the explicitness of the concept. What is more, the acronym for all of them is always the same, i.e., FEL. In Figure 1, we can see a graphic representation of FEL, which is equivalent to Community.

/   \
Fundamental Environment of language
Fig. 1

FEL is reminiscent of Peirce's sign. In this case, L would correspond to his representamen, P is equivalent to his interpretant, and T to his referent or object (Peirce 1972: 94). The broken line linking L to T indicates that there is no direct relation between language and the world, here represented by the T of territory, exactly as there is no direct relation between the representamen and the referent in Peirce's sign. Based upon Bandura, Wilhelm Trampe represents the "reciprocal determinism between behavior (V), Person (P) and environment (U)" as seen in figure 2, which is very similar to my Figure 1.

/    \
P = Person (=population, people)
V = Verhalten (behavior, language)
U = Umwelt (environment, territory)
Fig. 2

Still according to Trampe, Bandura "enlarges the classic formula of Lewin 1935 (Lebensraummodell V = f(P, U)", implying an interrelationship between V, P and U.                                                      

In summary, what the Fundamental Ecosystem of Language (FEL) shows is the common-sense idea that, in a prototypical situation, for there to be a language (L) there must be a people (P) living on a specific place or territory (T) that uses it. The equivalents of these concepts in ecology are behavior (interrelationship), population and biotope, respectively.

The totality of the individuals that make up the people (population), together with the actual and potential interactions among them constitute the Social Environment of Language (SEL). That is, SEL is the people held together by the rules of behavior, which include the language they have developed. Formally, we could say that SEL equals L+P, where P corresponds to the totality of its members qua social entities. Community is another name that could be given to SEL. The rules of behavior must be stored somewhere. As Haugen (1972: 325) says, "language exists only in the minds of its users". In this case we have the Mental Environment of Language (MEL). Here the rules of behavior formed by the population held together socially are stored and processed. Finally, the members of P must live in a specific territory, where they interact, not only among themselves but also with the physical environment. This physical or natural aspect of the environment constitutes the Natural Environment of Language (NEL). From an ontogenetic point of view, first comes the world (T), then plants and animals, among them people, and, finally, language.   

In addition to common sense, the idea of a major ecosystem or environment of language (FEL) as well as its component ecosystems SEL, MEL and NEL have something in common with the "dimensions" of Døør & Bang (1996: 23). My Social Environment of Language corresponds to their "socio-logic dimension" of language, a self-explanatory concept which needs no further clarification. My Mental Environment of Language is equivalent to their "ideo-logic dimension", if we consider the fact that they also use the term "mental" in defining it. My Natural Environment of Language is reminiscent of their "bio-logic dimension", since they also use the term "physical" for it.

Apparently, all concepts and principles of the ecological ecosystem are applicable to the ecosystems (or environments) of language. For instance, an ecosystem has no clearly defined limits. There are always transitions between any given ecosystem and the next. Its limits are defined by the investigator, who may demarcate a pond, a lake or the entire ecosphere as an ecosystem (Tansley 1935). The same happens with language ecosystems. Dialectologists and sociolinguists are well aware of the difficulty to demarcate where German ends and Dutch begins. One possible exception could be the Mental Ecosystem of Language because it is formed by the brain, which is entirely inside the skull. In the following three subsections I will investigate each of these language ecosystems in more detail.

The ecosystem of the language whose prepositions are the subject of this present paper, Kriol, is a highly heterogeneous community or FEL, namely, the country Guinea-Bissau. Its territory (T) of 36.125km2 was demarcated arbitrarily by the colonizers, ignoring the limits of the local ethnic groups. As a consequence, its population (P) of roughly over one million inhabitants consists of only 20% native speakers of Kriol. Approximately 60% learned it as L2, although most of them have it as their primary language today. That is, the total number of Kriol speakers amounts to roughly 80%. The remaining 20% have one of the sixteen ethnic languages as their mother language, mainly in rural areas, although almost all of them have at least a passive knowledge of Kriol. Kriol is spoken also in Casamance, situated to the south of Senegal. In other words, Kriol FEL stretches beyond Guinean borders, and most ethnic languages are also spoken in neighboring countries. To complicate the situation even more, the official language is Portuguese, which is not acquired by any Guinean of African ascendance as a mother language. For more details, see Couto (1994).

3.1. The Social Environment of Language

The Social Environment of Language (SEL) is the most conspicuous of the three so much so that Haugen (1972: 325) said that "the true environment of language is the society that uses it as one of its codes". Most present-day ecolinguists deal exclusively with it, as is the case with Fill (1993), Mühlhäusler (2003), Calvet (1999) as well as with all contributors to most of the collections of essays published to this date (see Fill & Mühlhäusler 2001, Fill, Penz & Trampe 2002). The only essays that could be said not to deal with some kind of social aspect of language are those geared towards philosophical aspects of the discipline. I say this not in reproach, but just to underline the fact that language is also related to other types of environment, as was hinted at, but not developed, in Fill's definition of ecolinguistics seen above.

An argument in favor of the idea that SEL is the most conspicuous language environment is the common sense conception of language mentioned above. If we could gauge the epistemological foundations of all linguists in the world, perhaps over 60% of them would consider language as mainly a social phenomenon. Mental theories like generative grammar are less common than social ones. All language theories that emphasize interaction privilege the social aspect (SEL). This is because, as Døør & Bang (1996: 23-24) state, "without a social praxis there would be no natural language nor linguistic sign nor text. A natural language is a cultural medium which stylizes and formalizes forms of social interactions". Even if language is stored and processed in the brain, everything in language begins in social interaction, and ends in it. In other words, it is social interaction that creates the rules that will make up the language. The role of the brain/mind is to store and process what happens in social interactions and in interactions of the members of P with the environment. Ferdinand de Saussure would seemingly subscribe to all of this. 

Among the subjects that have been investigated under this heading are, for example, environmental discourse, the discourse of polluters, anthropocentrism, ethnocentrism, androcentrism (sexism) in language as well as the ideas of development and bigness. Since the goal of occidental societies is to grow, to develop, big is good, whatever implies smallness is not good. In the same vein, ecolinguists defend language diversity together with biological diversity. This comprises minority languages, dialects and other language varieties. In summary, they defend all type of diversity in language.

The subjects that have been traditionally studied by sociolinguistics, such as multilingualism, bilingualism, language variation and language contact are also of interest to ecolinguists, as is implied in Haugen's text. In cases in which a state is confronted with many languages in its domain, it has to decide (impose) which language is the national, official or state language. This is the focus of language planning, to which Haugen himself dedicated many studies.

Individual experiences that remain in the individual, that is, those not shared with other individuals of the community disappear. They may persist in the individual during his life. However, when s/he dies, these experiences die with her/him. They do not form part of the language.

3.2. The Mental Environment of Language

The Mental Environment of Language (MEL) is the mind and the brain. The brain, or central nervous system, is divided in two hemispheres, the left and the right hemisphere. It is linked to the peripheral nervous system which connects the individual with the external world, i.e., the Natural Environment of Language. It is an irony that the most limited language environment is the least known. However, there is a reason for this. The brain is located inside the skull, therefore it is very difficult to access without injury. However, Pierre Broca (1824-1880) and Carl Wernicke (1848-1905) made important discoveries. Broca had come to the conclusion that there is left hemisphere dominance in language processing, which came to be called lateralization. Wernicke, in his work, discovered that sound images were located in the left temporal lobe, posterior to the primary auditory cortex. With the advent of noninvasive techniques such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), the path was open for further investigations on the relationship between language and the brain.

According to Sydney M. Lamb, our information system is "implemented mainly in the cerebral cortex and associated with matter, which provides cortico-cortical connections" that "includes the linguistic system together with conceptual, perceptual, and other systems" (Lamb 2000: 5). Still according to him, words are represented in the brain as nodes where several relations interconnect. In this case, "a node for a conceptual category seems to have connections to/from a large number of nodes representing its properties, both to/from other conceptual nodes and to/from other conceptual subsystems. For example, concepts for categories of visible objects need connections to nodes in the visual area, those for audible objects to/from auditory nodes, and so forth. Taking the concept Ccat, for example, we have visual connections comprising what cats look like, auditory connections for the 'meow' and other sounds made by a cat, tactile connections for what a cat feels like to the touch; as well as connections to other concepts representing information about cats in the information system of the person in whose system these connections have been formed" (p. 6).

When we hear the word "banana", for example, the image of banana as well as the sounds that compose it are immediately activated (maximum activation). However, other concepts and sequences of sounds are secondarily activated, as is the case with "banana oil" and "banana split" (high activation). Words such as "yellow" and "mango" are activated to a lower degree (medium activation). Words like "bank", "sky" and "Neptune" remain inactivated (França 2005). This reminds us of Saussure's paradigmatic relations.

There are other known interrelationships between language and its mental environment. For example, we know that the arcuate fasciculus connects phonological recognition to phonological production. There are also some other grammatical facts that are known, for example, in relation to which part of the brain is activated during the production and perception of phrases.

What happens in social interaction can only be continued and become part of a group's language if it is stored in the brains of the members of P. Only in this case can it be reused and recycled. For Chomsky, for instance, the brain is the real locus of language because it is stored and processed in a complex network of neural connections (synapses). However, everything in the interrelationships between each member of P and the environment, and in turn, with other members of P as well as with the physical environment precedes the mental process. 

As far as I know, there is no investigation of this language environment from the perspective of a language-and-environment approach. However, psycholinguistics, neurolinguistics and connectionism have produced research that can be appropriated by ecolinguistics, as Haugen himself had foreseen. Even Lamb's neurocognitive linguistics (previously known as stratificational grammar) has contributed to the knowledge of what goes on when we talk or when we hear someone's speech. In Makkai (1993) there is a discussion of this theory of language.  

I do not have much information about what happens with spatial concepts in the Mental Environment of Language (MEL), the brain/mind. However, it is known that the acquisition of prepositions by children generally proceeds from "in" to "on" to "under". Even "next to" precede "before/after". Sometimes the latter is preceded by "between". What is important here is the order of acquisition "in/out > on/under > before/after". This implies further evidence in favor of what is said below regarding the Ecology of Spatial Relations. It has been shown that a lexical item like a preposition emerges when the child has the equivalent experience with the world, although there is disagreement on this point.

3.3. The Natural Environment of Language 

The Natural Environment of Language (NEL) is the least investigated by linguistics, if at all, among the three that constitute the Fundamental Environment of Language. However, in other areas as philosophy it has been the focus of some investigation, beginning at least with the Greeks, as already mentioned in the Introduction. In this case, one studies the relation between word and thing in what is generally called designation, denotation, reference etc. However, the relation between language and physical environment is not confined to the lexicon. According to Ogden & Richards (1972: 52), for Heraclitus "the structure of human speech reflects the structure of the universe".

Closer to present times, John Haimain defends similar ideas. He begins by reproducing Joseph Greenberg's assertion that "the order of elements in language parallels that  in physical experience or the order of knowledge", as in Veni, vidi, vici (Haiman 1980: 528). He concludes by saying that "since the transformational revolution, it has been claimed that the structure of languages reflects the structure of THOUGHT, and that its study provides 'a window on the mind'. In arguing, as I have done, for the iconicity of grammar in general, I contend that the structure of thought in its turn reflects the structure of REALITY to a far greater extent than it is now fashionable to recognize. Ultimately, I believe that many (if not all) of the formal universals of syntax which now engage the attention of most syntactic theoreticians will, if valid, be found to reflect properties of the world rather than properties of the mind per se". He finishes with Bertrand Russel's claim "that 'partly by means of the study of syntax, we can arrive at a considerable knowledge of the structure of the world'" (Haiman 1980: 537).

There are at least two ways of regarding the relationship between language and environment. The first one goes back to Hermogenes, according to Plato's Cratylus, as seen above. More recently, it is represented by a tradition that goes from, for example, Hamman (1730-1788), through Herder (1744-1803) and Humboldt (1767-1788) to Weisgerber, to mention just a few of them. According to this tradition, language lies between us and the world, as we can see in Humboldt's Weltbild (world view) and in Weisgerber's Zwischenwelt (interworld). It is also represented in a tradition that begins with Boas (1858-1942), goes through Sapir (1884-1939) and Whorf (1897-1941), culminating in the famous Sapir-Whorf Hypothesis. In some versions moreover, this hypothesis says that there is a symmetry between language and what it represents (Wittgenstein 1968), with language between us and the world. Helbig (1975: 124) represents this view graphically as can be seen in Figure 3 (T stands for physical world in general).

Fig. 3

According to this view, the world is constructed by language, so that we only have access to it through language. We would see in the world only what our language shows us. However, as Schaff (1974) says, this view is idealistic and mystical, although it has some truth to it, namely that every language categorizes the world differently.

As the Norwegian philosopher Arne Naess says, "in spontaneous experiences we have direct access to what is real". According to him, "we do not construct things. We construct concepts of things" (Naess 1997: 2), namely words, language. This is a good starting point to discuss the opposite of the previous interpretation of the relation between language and the natural world. In fact, it is the people (P) that lie between language and the world. For this reason, Mufwene (2001) considers L a parasite of P, in the biological sense. This concept may be represented as seen in Figure 4, which is nothing more than a linearization of Figure 1.     

Fig. 4

According to this second conception, we have direct access to the world. In fact, we build language with material taken from it. However, after formed, language acquires a relative autonomy vis-à-vis the world. After having the resources for forming sentences like "Voiceless stop phonemes occur frequently" speakers can also construct sentences like "Colorless green ideas sleep furiously". Both are grammatical. The morphological grammar of English admits not only "permit, percept, perceive", on the one hand, and "commit, concept, conceive", on the other, together with "deceive", but also "*deceit, *demit" (* = inactivated), so that the number of inactivated words is very high. However, they are built by reusing or recycling already existing morphemes. In phonology there are inactivated syllables, as is the case with /flEs/ in Portuguese. It is well-formed precisely as /frEs/, which occurs in the word "fresta" (cleft, gap), but /flEs/ is not yet activated, i.e., it is not yet used in any known word of the language. Even monomorphemic lexemes can contribute to this relative independence of language. One case is their assuming new meanings, which enriches the referential potential of the language. Among the strategies used for this purpose are metonymies, metaphors, polysemies etc. Regarding the strangeness of some inactivated constructions, there is an extensive discussion in Couto (2006).

It is this relative autonomy of language in relation to the world that gives the impression that language is independent of us (God-given?) and greater than the world, having the power to create it. It could be like a mechanical mirror-image of the world, as can be seen in the metaphysical tradition as well as in, say, Wittgenstein's Tractatus logico-philosophicus. 

The relation L-NEL can be examined from other points of view. One of them is verbal interactions between individuals belonging to the same language community. These interactions take place at a specific place. When individuals meet at some physical place, the simple fact that they are together generally leads them to interact. In some sense, interaction (and verbal interaction) is determined by co-presence in space. Language emerges out of interactions. This can be clearly seen in the formation of a creole language, where there is the convergence and close interaction of people (P) from different cultural backgrounds (language, territory, customs etc.) that may lead to the emergence of a new language. Interaction is one of the most important forms of social praxis.

Mental concepts exist due to the existence of physical stimuli in the external world, in the NEL, as can be seen in the ecology of spatial relations, below. I do not have specific information concerning spatial prepositions, but I dare to hypothesize that in mountainous regions concepts such as "before/after", or "high/low", might be more prominent than, say, "on/under". In fact, there is a parallel case mentioned by Sapir, who, in his famous "Language and environment", mentions "the use in Kwakiutl and Nootka, for instance, of local suffixes defining activities as taking place on the beach, rocks, or sea" (Sapir 2001: 21).

Of the three language environments that are part of FEL, it is the Natural Environment that is directly related to the Ecology of Spatial Relations. In this case, everything that is said in the next section applies here as well. In my investigations, I privilege the relationships between language and the physical or natural environment.

3.4. Interrelationships between the Social, the Mental and the Natural Environment of Language
The preceding discussion about the Social, the Mental and the Natural Environment of Language in three different subsections may give the impression that they are independent from one another. There is nothing further from the truth since the concepts and principles that constitute the epistemological basis of the whole discussion are taken from ecology, the idea that everything is related to everything is implicit everywhere. As we saw in the first part of the present section, the central concept of ecology is the ecosystem. Ecology tells us that an ecosystem consists of the interrelationships among organisms and between them and their environment. Further, the vitality of an ecosystem depends on its diversity, especially biodiversity. Here we have some of the most important concepts of ecology that social sciences can use, namely, a) ecosystem, b) interrelationships, c) diversity. The interrelationships imply movement, dynamicity, i.e., d) evolution, which constitutes a fourth concept.

In the ecosystem(s) of language, or language environments, the same thing happens. Within the general ecosystem of language, the Fundamental Environment of Language, there is a close interrelationship among the three sub-environments: SEL, MEL and NEL.

Let me illustrate this with the case of the prepositions in (2), below, borrowed recently from Portuguese. The first time each of them was used, it was proffered by an individual and directed to another individual of the Guinean community. This was a social act. However, the information for this social act started in the brain of the speaker, where it was associated with the respective meaning. From the brain it was taken to the speech organs, from where it was taken to the ear of the hearer. The hearer, in turn, processed it in her/his brain, possibly giving a response to the speaker. The process inside each individual belongs to MEL. However, it was shared with at least another individual, the hearer. This means that SEL was involved as well. Further, the whole interaction took place somewhere in the territory of the individuals in question, which means that NEL was also active.

Everything begins in NEL. It is the encounter of two members of the community at a specific place in the territory that enhances the communicative interaction. Whenever two people see themselves together in a determined space, they interact. The interaction begins as a psychic process and goes on as a social process. In summary, everything begins in NEL, passes through MEL and is then sanctioned by SEL. Any linguistic innovation emerges in the individual (MEL), where it may be stored. The storage continues only if it is shared with one or more individuals of the community (SEL). The whole process is possible only in a specific setting of the territory where community members live. That is to say, there is a constant interplay of the three language environments, forming a cycle, where it is difficult to discern where is the beginning and where is the end. The beginning is the end and vice-versa. All these interrelationships take place almost simultaneously.    

4. The Ecology of Spatial Relations
Investigations of the relationships between language and the world have been restricted almost exclusively to the lexicon, as can be seen in the philosophical tradition (relation word-thing) mentioned above as well as in Sapir (1963). According to this author, "It is the vocabulary of a language that most clearly reflects the physical and social environment of its speakers" (Sapir 1963: 90). However, there are some exceptions to the idea that only the lexicon is related to the environment, as is the case with Heraclitus' and John Haiman's ideas, as can be observed elsewhere in this present paper. However, even when we restrict the study of the relationship between language and environment to the lexical component of language, in general only content words, i.e. names of things existing outside of language, are considered. What I intend to do here is investigate these relationships as they are implemented in specific kind of grammatical words, namely, prepositions. 

In order to do this, let's see what might be called Ecology of Spatial Relations. The idea in itself goes back at least to as far as 1668. In that year, John Wilkins represented these relationships in a two-dimensional space. He put an observer in front of a point of reference consisting of two concentric circles. First of all, he showed a "downwards" arrow, whose result is "beneath", as well as an "upwards" arrow, whose result is "above". Both relationships include the observer himself. Then come the following relations: a) "within" versus "without", b) "into" versus "out of", c) "upon" x "under", d) "on this side" x "beyond", e) "at" x "from". The following relations in Wilkins' scheme are not dichotomous: f) "about", g) "over". Some positions in his representation are less clear. For example h) "below" lies between the observer and the circle, but a little below the diameter of the circle, not under it. Beyond the central circle, again from the observer’s point of view, Wilkins notes: i) "against", j) "through", k) "besides". Outside of the larger circle, he puts l) "to" x "off". Finally, we have m) "after" (to the back of the observer [sic!]) versus "before" (in front of him, i.e., between him and the point of reference).

As can easily be seen, there are several inconsistencies in Wilkins' attempt at an "ecology of spatial relations". The most conspicuous one is the positioning of "after": it ought to have been placed at the opposite side of the circle in relation to the observer, not behind the latter. Further, "against", "through", "besides" as well as "without" are placed after the point of reference. The only preposition that belongs here is "without". Finally, the position indicated by the preposition par excellence, namely, "in" and its counterpart "out", are absent from his scheme. Perhaps this is due to the fact that Wilkins deals only with movement prepositions, as Eco (2002: 319) noted. However, we should not be too critical of him. His study was probably the first attempt to do this in the history of the study of language.

Bernard Pottier also dealt with the question, although his representation is only one-dimensional. Drawing on Gustave Guillaume's ideas, Bernard Pottier suggests what could be represented as can be seen in figure 6 (see Pottier 1962, 1969). Everything is before or after the point of reference. He includes not only prepositions.
-------->| -------->
to     |   from
until  |   since
before|     after
under, beneath| on, over, above
…. |  and
…… |  more
Fig. 5

 In my master’s thesis, I proposed a three-dimensional model of spatial relations, unaware at the time of Wilkin’s representation (Couto 1973: 45). In the nineties, I worked more on this idea in a book directed specifically to Kriol. This second version of the representation is reproduced in Figure 7 (Couto 1994: 122). The English equivalents of the Portuguese prepositions are roughly the following: em 'in', ante 'before', após 'after', sobre 'on', sob 'under', à direita de 'to the right of', à esquerda de 'to the left of').

                                                                       -------------------  ß à esquerda de
            Observer à                             antes   |           em       | após
                                               à direita deà  |                       |
                                                                                     Fig. 6

The central point of this ecology is insideness (interiority). This is probably due to the experience of our ancestors who, for example, noted the fact that a seed or a pit is inside the fruit, the stone of a fruit. This location is independent of an observer. The seed is objectively there. Therefore, interiority can be considered the unmarked spatial relation, together with its opposite, outsideness (exteriority). The simple preposition that represents insideness is the unmarked preposition. This position is independent of the dimensions of horizontality and verticality, which are important for the remaining positions in figure 7. It is true that Lyons (1977: 690) says that "verticality is physically and psychologically the most salient of the spatial dimensions". However, a little below he adds that "the notion of containment, or interiority, is obviously a very basic notion", so that "there may be grounds for introducing it into the analysis of the meaning of such prepositions as 'above', 'below' etc." (p. 699). The first assertion obtains only when "we take orientation into account" (p. 702). Vandeloise (1991: 34) adds that "the extralinguistic container/contained relation [is] practiced by the child well before language is fully acquired". 

The relation of superiority (on-ness) and its opposite inferiority (underness) are located on the vertical axis. They are independent of an observer. On the horizontal axis, we have anteriority (beforeness) and posteriority (afterness), which also depend on an observer. This relation is more marked than the preceding ones. In any case, I think that we could reasonably consider them with the preceding ones as representing prototypical spatial relations, at least as far as prepositions are concerned so much so that practically all the Romance languages, and English, have a nominal designation for each of the positions at stake. In French, for example, they are: intériorité/extériorité, antériorité/postériorité, supériorité/infériorité. Finally, we have the horizontal dimension "to the right/left of". In most European languages, there is neither a simple lexeme for them, nor a nominal designation. "Dextérité/sinistrité" do not seem to be usual words in the language. 

The remaining spatial relations would not be prototypical, in the sense of prototype theory (Rosch 1975, Coleman & Kay 1981, Bärenfänger 2005). In one sense or another, they can be derived from the prototypical ones. This is the case with movement prepositions, as those represented by Wilkins. According to some authors, all remaining relations, and the prepositions that represent them, are in some way derived from spatial ones.

I would like to observe that the relations seen in Figure 7 do not exhaust possible relations within the ecology of spatial relations. For example, along the vertical dimension, there are positions such as "high/low". Along the horizontal axis, there is also deixis like "near/far". Other possibilities are dimensions like "broad/narrow", "long/short" and so on (see Greimas 1966: 33).

5. Kriol prepositions: an overview
According to Kihm (1994: 67), Kriol has simple prepositions, as those in (1a), and complex prepositions, as those in (1b).

(a) a 'to', di 'of, from', entremetadi 'between', kontra 'against', ku 'with', na 'in, at, on, to', pa 'for, by, toward', sin 'without', suma 'as', te 'until'

(b) antis di 'before', (na) banda di 'about, close to', ba(s) di 'under, below', dentru di 'inside', dianti di 'before, in front of', di(s)pus di 'after', disna di 'since', filadu di 'in front of, across', fora di 'outside, out of', juntu di 'near', lungu di 'along', lunju di 'far from', na metadi di 'between', na roda di (around), pa bia di 'because of', riba di 'on, above', pertu di 'near', te 'until',  tras di 'behind'

Besides these prepositions, there are some that are clearly late borrowings from Portuguese. Some of them are reproduced in (2).

(2) entri 'between', anti 'before', sugundu 'according to', cunformu 'according to'

There are also some constructions that, without being exactly headed by prepositions, are equivalent to prepositional constructions, referring especially to spatial relations. In (3) we can see three examples.

(a) Maria bai na si tras 'Mary goes at his back, Mary goes behind him'
(b) Jon sta na si dianti 'John is in his front, Mary is in front of him'
(c) Jugude bua na si ladu 'the vulture flew at his side, the vulture flew along him'

Alain Kihm agrees with the idea that the Kriol preposition par excellence is the one that encodes meanings such as "be-inside", "be-on-the-surface", "be-at-a-location". In other words, the preposition that encodes interiority (na). Kihm’s examples are reproduced in (4).

(a) i sta na kuartu 's/he is in the room'
(b) i sta na mesa 'it is on the table'
(c) i sta na fera 's/he is at the market'

Let's take a look at the frequency of occurrence of Kriol prepositions in order to have an idea of their vitality in the language. I counted simple and complex prepositions in nine folk-tales and in one long speech by a politician. The former are basilectal, whereas the latter are mesolectal to acrolectal. The results are shown in Table 1 (disna 'since' did not occur in these texts).

dentru di
riba di
pertu di
antis di
na metadi di
dipus di
pabia di
bas di
tras di
na roda di
pa ladu di

Table 1

In Table 1 we can see that the most frequent preposition is di. However, this is certainly due to its occurrence in idioms such as manga di 'a lot of' (9 times), gosta di 'to like', dibi di 'ought', di maña/tardi/noti 'in the morning/afternoon, at night'. In this case, it is probable that the most frequent preposition is na. It also occurs 47 times combined with other morphemes, as in n'el (in it), n'unde/n'de ("in where"), although di may also combine, as in d'el/d'elis (of /herhim, her/his; of them, their), but only 9 times. If we count the preverbal particle na, which indicates nonpunctual action in Bickerton's (1981) terminology, as in Jon i na fuma 'John [he] is smoking, there would be no doubt that na is much more frequent than di. In fact, in 1,176 occurrences, na was verbal particle 690 times (58,67%), and preposition 486 times (41,32%). As Claire Lefebvre said about similar morphemes of Haitian Creole, na is a multifunctional morpheme (Lefebvre 1998). Both come from the combination of the Portuguese preposition "em" (in, at, on) combined with the feminine definite article "a" (the). One would be preposition with noun, the other "preposition" with verbs. 

The third and fourth most frequent prepositions (ku, pa) may also combine with other morphemes, but to a lesser degree. In my statistics, ku appears 6 times in k'el (with him/her/it) and one time in k'elis (with them). As for pa, it appears in par maña 'in the morning' (1 time) and par'el (for him/her/it) (2 times).

There are other peculiarities of Kriol prepositions that are worth mentioning. One of them is the fact that riba di (on, above) may also occur as na riba di. Another one is the relatively high number of complex prepositions borrowed from Portuguese, especially in acrolectal discourse. In the one just mentioned, I found: ao lado d'el (at his/her/its side) and pa/por exemplo (for example). These examples add five more occurrences of pa. Finally, in the acrolectal text I had one occurrence of na frenti (in front).

It should also be noted that the prepositions ku and pa function as complementizers. The first occurs in rapas ku ka misti tarbaja (boy, who, NEG, want, work 'a boy who does not want to work'); the second, in N misti pa bu bai (I want for/to you go 'I want you to go').

6. The Ecology of Spatial Relations in Kriol prepositions
In this section I will examine Kriol prepositions in light of the Ecology of Spatial Relations, as represented in figure 7, and illustrated with Portuguese prepositions. Looking at the list of prepositions in (1a) above, we can see that the only equivalent to the simple prepositions of Portuguese in Kriol is na. In order to fill the remaining slots we have to resort to complex prepositions (5).

in = na, dentru di (144)
on = riba di (7)
before = antis di (3)
after = dipus di, tras di (2)
under = bas di (1)

Regarding "to the left of" and "to the right of", I did not find any equivalents for them in my corpus. This is an indication that they might not have been so important for Kriol speakers. 

We have already seen that acquisition studies of English prepositions as well of other languages have shown that first children learn "in". Next, come "on", "under", "next to", in this order. In general, the next to be acquired are "next to" and "in back of/in front of". If we compare theses findings with the statistics of occurrence of Kriol prepositions in Table 1 reproduced in (5), we see that they coincide at least in part. The two most frequent prepositions are na (in) and riba di (on), in this order. However, there is a discrepancy, in that bas di (under) is the less frequent, which could be due to the paucity of data I scanned. However, the number of occurrences of the remaining two confirms the order of acquisition, that is, after na (in) and riba di (on) come antis di (before) and dipus di (after).   

As already noted, out of the above Kriol names for the spatial relations in question, na is not only a simple lexeme, a monomorphemic lexical item, but also a monosyllable. All the others are complex prepositions. All of them consist of an adverb or a noun followed by the preposition di (see section 12). This is a further argument in favor of the assertion that na is the preposition par excellence. It exists in practically all Portuguese and Spanish based creoles as well as in non-creole varieties of these languages such as Puerto Rican and Cuban Spanish. It has been reported to exist even in creoles having other European lexifier languages such as Sranan, St. Lucian, Trinidadian and, maybe, in Virgin Islands Creole Dutch, Jamaican and Krio (Thompson 1961: 112). In some popular/colloquial variants of Brazilian Portuguese, the form ni may occur, instead of the usual em (in). Probably, there is a relation between it and Kriol na.

An additional argument to confirm the thesis that na is the preposition par excellence is the fact that it can fill practically all positions in Figure 7, as shown in the examples in (6). The strongest argument, however, is the fact that it encodes the core position in the Ecology of Spatial Relations, i.e. interiority.

(a) lebri sta na si koba 'the rabbit is in her/his burrow' (in, inside)
(b) lebri sai na si koba 'the rabbit came in [out of] its burrow' (out of, from inside)
(c) libru sta na mesa 'the book is on the table' (on)
(d)  Jon bai na bu kasa 'John went to your house' (to, before)
(e) N bai na bu kasa 'I went in your house' (to)

With respect to the positions "to the right of/to the left of", the closest Kriol equivalent I was able to find is na ladu di (on the side of), exemplified in (3c). It indicates laterality in general, as is also the case with the others examples in (3).

The dimension antis di - na - dipus di is normally horizontal. However, depending on the position of the observer and of the thing observed, it can be used to indicate other directions as well. For example, in an upward direction, we could say that a squirrel is before a bird on the trunk of the tree, if we are looking from the ground. If we are looking from a helicopter or from a higher branch of the tree, we could say that the squirrel is after the bird.

The relation of the Social Environment of Language vis à vis this paper’s focus is only indirect. Perhaps we could adduce here some social uses of certain positions in space. In Brazilian Portuguese, "estar por dentro" (to be inside) means to know what is going on, "estar por fora" (to be outside) means the contrary.

In English, "to be in" is something desirable, whereas "to be out" is not. In Christian tradition, heaven is above us, whereas hell is below us. Therefore, "on/over/above" is good, "under/below/beneath" is bad. In Brazilian culture, for instance, to be above someone or some subject is to dominate it, in all senses of the word; to be below means the opposite. In Romance languages the word for hell is derived from "infernus", a word that has to do with "inferior". Apparently, this opposition has the same connotation in Kriol. There is a tale entitled "Gera di gintis di riba ku gintis di bas", that is, war of the people from above (birds) with the people from below (other animals). It is the first group that wins the war. In occidental societies, to be in front of, to be ahead is better than to be in the back of.

The Portuguese word for the opposite of the concept of "to be ahead of" is "atrasado", derived from "atrás" (in the back). It is the contrary of "avançado" (advanced), vanguard. The distinction right/left also has political connotations. A leftist considers himself as more conscious, democratic and liberal than the rightists. This originated in the French Revolution. In the meetings of the Legislative Assembly (1791-1792), the aristocracy, the jacobines, were seated on the right, whereas the democrats, girondines, sat on the left.

7. Other spatial prepositions in Kriol
As can be seen in Table 1, there are other prepositions in Kriol that indicate spatial relations, in addition to those appearing in (5). Among them we have entremetadi (between), na metadi di (between), na roda di (along), (na) banda di (about, close to), filadu di (in front of, across), fora di (outside, out of), juntu di (near), lungu di (along), lunju di (far from) and pertu di (near), all of them indicating static relations, like those in (5). To these spatial prepositions we must add movement prepositions as those in Figure 8.

antis di
dedi, disna (di)
dipus di
bas di 
riba di
fora di
na, dentru di
                                                                             Fig. 7

The prepositions in Figure 7 are used typically with movement verbs. They are the prepositions that prototypically belong to Pottier's scheme, whose arrows suggest directionality. In spite of this traditional way of regarding them, I would like to suggest that, what indicate movement are generally the verbs to which they are attached, not the prepositions per se. What happens is that they are compatible with this kind of verb.

Here the subspecification of movement is included, namely, directionality, source, path and goal (Bennett 1975). Anyway, I will continue to use the term "movement preposition" and its derivatives, but keeping this observation in mind. The prototypical movement prepositions in English are "to" (-->|) and "from" (|-->). Their equivalents in Kriol are pa and di, respectively. The Kriol forms te (up to, until) and desdi/disna (since) could also be so considered.

According to Pottier's model, all prepositions, as well as some adverbs and other kinds of words, should fit along one or the other side of the vertical line. That the prepositions pa/di and te/dedi(disna di) belong to the positions they fill in Figure 8 is indisputable. Even antis di/dipus di do not present any problem. Disagreement begins regarding such forms as bas di/riba di, fora di, na/dentru di.

Pottier claims that the left side of the vertical line indicates afference, whereas the right side indicates deference. He also says that the first "proposes", indicates approximation, and the second "disposes", indicates distancing. With this additional information, it becomes easier to accept bas di and fora di at the left and riba di and dentru di at the right side of the figure. As to na, its placement on the right side is not so obvious, especially if we consider constructions such as those in (18), where it indicates goal. Apparently, it should be placed at the neutral point, indicated by the vertical line.

8. Apparently nonspatial prepositions in Kriol
Some prepositions seem not to be spatial, at least apparently. Among them we could mention ku (with), sin (without), suma/sima (as, like), kontra (against) pa bia di (because of). However, all of them belong to one or the other side of Pottier's model, as in figure 8.
sima, suma
Pabia di
Fig. 8

In terms of ku, Pottier (1962: 279) says that its Latin equivalent is a semantic variant of ad (to, towards), a typically spatial preposition. On its English equivalent, John Lyons says "that what John is with Peter means is John is where Peter is" (Lyons 1977: 693). From a logical point of view, ku indicates conjunction, that is to say, simultaneity of two beings in space, or in time, as it were. Another argument in favor of this interpretation is the fact that in Kriol ku has the function of English "and" as well. This happens in many titles of folk-tales, like storia di lubu ku lebri (tale of the wolf and the rabbit). As to sin (without), it is the diametrical opposite of ku, therefore it revolves around the same axis. In other words, it has spatial connotations.

If Pottier is right in placing all prepositions in the "to" or the "from" side of his model, all prepositions should have a spatial connotation. After this observation, let's go to suma and sima as exemplified in (7).

(a) bu firma suma po 'you are standing like a tree'
(b) sima fonti na kamiñu 'like a fountain near the road'

To begin with, suma and sima seem to be simply two phonological variants of the same preposition. In fact, there are many alternations i/u in the language. The conjunctions si (if)  and ki (that), for example, have the variants su and ku, respectively. To confirm this interpretation, we could add the fact that sima is very rare. It did not occur in my corpus. What is more, out of all the texts I have seen (ca. 11), it occurred two or three times in only two of them.

Some authors consider kontra a preposition. However, in my corpus it occurred as verb (8a), adverb (8b) and conjunction (8c), preceded by si (if), but not as a preposition. The authors that considered it a preposition may have seen it in a decreolized variety of Kriol. In this case, it would be a borrowing from Portuguese contra (against).

(a) bu ba kontra gora ku un algin 'you will now meet somebody'
(b) kontra kau firia omi bin sai di si kasa 'when it became cold, the man came out of his house'
(c) si kontra bu papia son, ami N fugial 'if you keep talking, I'll shoot you'

Practically all prepositions can also indicate notion, in Pottier's sense. Some of them, as well as the remaining prepositions in Figure 9, will be discussed in the next section.

In order to discuss pa bia di, let us start with its English translation "because of". This preposition contains in itself the form "cause", which is something that precedes an effect. In this case, it would contain some temporality and, as we have seen, temporal relations are reducible to spatial ones, as it were.

In their origin, all of these prepositions may have been squarely spatial, as claimed by Pottier. Not only temporal relations, but also "pure" (or "logical") relations. In summary, it is highly probable that genetically, most prepositions emerge as spatial prepositions, subsequently they may be used with temporal connotations. Later on, they could be used in the domain of notion, i.e., connecting abstract concepts. The genetic order would thus be: space > time > notion (see Pottier 1962).  I shall return to this in section 13.

9. Temporal prepositions in Kriol
The horizontal dimension "before-in-after" in figure 7 can be seen as a subset of the Ecology of Spatial Relations that indicates linear relations. In English there are other sets of prepositions in the same case, as, for instance, "to-at-from", "onto-on-off" and "into-in-out of". Since the occidental Christian tradition sees time as a line, the prepositions encoding these dimensions could, in principle, indicate temporal relations as well. In Figure 10 we can see the equivalents of these spatial prepositions in Kriol. In (9), they are inserted in phrases.                                                       

antis di
dipus di
                                                                                Fig. 9                          

(a) i kume antis di si dunu 'he ate before his master'
(b) Ña pape ku bu pape na si tempu ba 'my father with your father in their time'
(d) N bai dipus di Bisenti 'I'm going after Vincent'

Concerning na, it should be noted, however, that in my data it occurred in temporal construction only before time words such as tenpu (time) dia (day), anu (year), as  shown by the examples mentioned at the end of section 12. Even the other two occurred in very few phrases.  

These are spatial prepositions used also in temporal function. But, in Kriol there is at least one preposition that is used only to indicate temporal relation, namely disna (di), as in the examples in (10).

(a) N sta li disna k'u bai 'I am here since you went away'
(b) N ka kume nada disna di parmaña 'I did not eat anything since the morning'

Although dedi, which has the phonological variant desdi, can also occur with this meaning, disna is specific for this meaning. In other words, disna is used only temporally (11a), whereas dedi can be used spatially (11b) and temporally (11c). Although the examples found are not many, it  should be noted that disna contains the idea of "for a long time". 

(a) N ka kume nada dedi aonti 'I did not eat anything since yesterday'
(b) bu ta kumsa cora desdi kanpada di kasa 'You began to cry since the straw house'
(c) i ianda dedi parmaña sedu 's/he/it walked since early in the morning'

I would like to remember that in some dialects of Brazilian Portuguese we can find the form desna, with roughly the same meaning. Apparently, Kriol disna comes from it. The Portuguese form, in turn, probably would come from desde 'since, from'+ na (na = em [in]+ a [def. art. fem]). Unfortunately, I could not find any information about this probable etymology. 

The only temporal uses of di I was able to find occurred in fixed expressions like di dia (during the day), di noti (by night) and di menoti (at midnight). I did not find any use of it in constructions like "he went from to Bissau to Bolama".

In regard to te, it indicates a temporal relation, as shown in (12).

(12) Jugude ka ojal mas te aos 'the vulture has not found him/her until today'

Bennett (1975) recognizes a unified underlying meaning in each preposition. This underlying meaning can be put to spatial and temporal uses. In the following section, we will see that that there is the notional use too, not considered by Bennett.

In summary, there is one preposition used to indicate only a temporal relation. It is probable that its origin  was spatial. Unfortunately, I cannot substantiate this claim. However, if the proposed etymology is right, the spatial origin could be retrieved from desde (from) as well as from na.

10. Notional prepositions
According to some investigators, especially generativists, there are "pure", abstract relations. These relations would not have anything to do with either spatiality or temporality. Among the prepositions that could have such a function, we could mention those in (13).

(a) storia di kasamenti di fiju di reglu 'story of the marriage of the king's daughter'
(d) e dibi di ciga 'they should arrive'
(c) N misti papia ku bo 'I want to talk to you'

The first two occurrences of di in (13a) are notional, i.e. nonspatial and nontemporal. The third is genitival. However, spatiality is latent in all of them. The "story" revolves around (a spatial concept) "marriage", and "marriage" belongs to the "king's daughter". Genitivity indicates that the "daughter" somehow comes from the father, including genetically. Her origin lies, materially, in him. As a rule, 'X has p' can be analyzed as 'p is at X' (Lyons 1985: 722-724). The same could be said of 'p is of X'. Something similar could also be said of ku, as we have seen.

Pottier says that all prepositions can, in principle, be used in three domains, namely, spatial, temporal, notional. The latter covers what we could call "pure", "abstract" prepositions. Pottier's theory presupposes that even these are underlyingly spatial/temporal. Or, more precisely, they are reducible to spatio-temporality. Let me begin with na, as illustrated in (14).

(14) kila ka sta na si planu 'that was not in his plan'

Here, na is clearly used in a notional, i.e., abstract sense. Therefore, there is no direct spatiality nor temporality in it, at least apparently. However, as the English gloss suggests, we could interpret it as "location in the plan".

In (15) we have both di and ku, in the same sentence.  

(15) Sara papia di Jon ku Maria 'Sarah talks about John with Mary'

Both prepositions are used here notionally, that is, neither spatially nor temporally. However, the talk of Sarah had John as its base, its subject came from him.  Further, she talked to Mary, that is, she "sent" her talk to Mary. We could say, paraphrasing John Lyons, that she talked about John where Mary was. For similar arguments, see the example of serial verb in (20). 

All relations are reducible to spatial relations, not only temporal relations, but also "pure", "logical" or notional relations. In their origins, all of them may have been squarely spatial. Thus, when we talk "about" something, we are in the end talking "around" it. In Portuguese, this meaning is rendered by the same preposition that translates English "on", namely "sobre" (o livro está sobre a mesa 'the book is on the table'; eu falo sobre futebol 'I'm talking about soccer'). 

11. Some apparently strange uses of Kriol spatial prepositions
There are some uses of local prepositions in Kriol that seem a little strange to speakers of at least some European languages. However, when we analyze these uses in light of the Ecology of Spatial Relations, and keep in mind that natura non facit saltus (nature does not jump), we see that they are not strange at all. Let us start with the examples in (16).

(a)  i sai na si koba 's/he came out in her/his burrow' (why not "i sai di si koba"?) (in = of)
(b)  bajuda sai na kumbu di po 'the girl came out in the hole in the tree' (in = of)
(c)  i tira si fiju na dentru di galiña 's/he took his/her child in the hen' (in = from)
(d)  bó sai na matu 'you came out in the jungle' (in = of)
(e)  iagu ta sai na uju 'water began to come out in his/her eyes' (in = of)

In these examples, na is used with the meaning of "out of". In fact, na is static, it indicates location and, at most, goal, but not source. How could we explain this apparently strange use of na? Let's see what Derek Bickerton says on similar relations that take place elsewhere. Drawing on the acquisitionist Eve Clark, he presented the four relationships seen in Figure 10.

Figure 10

Bickerton also says that Eve Clark "reviewed in which ways these relationships are presented across a sample of fifty-odd languages". Bickerton himself applied the same principle to the framework of creole articles. In the case of Kriol, we could say that the morpheme that indicates possession also indicates existence, as shown in (9). The Portuguese translation shows that in Portuguese the same phenomenon can be observed.

(a) Jon tene libru 'John has a book; (a') João tem água
(b) i ka tene libru li 'there's no book here'; (b') não tem livro aqui

Bickerton concludes that "no language can use the same morpheme to express any noncontiguous relationships (i.e., location and possession, or existence and ownership) unless the same morpheme is also used to express one of the intervening relationships" (Bickerton 1981: 244-245).

In Kriol, the preposition encoding the position of interiority also encodes the English idea of "at", that is, a point in a line. Since it is adjacent to both "before/to" and "after/from", it could, in principle, replace both relations, as well as any one of the remaining positions in the ecology of spatial relations in (6). In this case, it is not so strange to have constructions such as those in (18). In Latin (18a) and Brazilian Portuguese (18b), its equivalents may indicate the same as English "to".

(a) eo in Romam 'I go to Rome'
(b) Eu vou lá em Roma 'I go to Rome'

Even the complex expression of location in the interior of, namely, dentru di, can be used in this function, as shown in (19).

(a) tuga sta dentru di no tera 'the Portuguese are inside our land / the Port. are in our land' 
(b) no tira tuga dentro di no tera 'we take the Portuguese inside our land' / 'we drove the P. out of our land'

In (19a), the use of dentru di  replacing the expected na sounds strange to our ears because the Portuguese were not "inside" Guinean's land, but "in/on/at" it. In this case, na would be expected, because of its polyssemy. In this example, the relation indicated by the preposition is static. In (19b), however, dentru di is used in the same sense as na in (16), namely, for "out of", with a movement verb.

David Benett says that "to the church” would be more accurately reflected if we could say in English *to at the church" (Bennett 1975: 18). The uses of na in (16) would be more accurately reflected if we could say in Kriol *di na koba (from in the burrow). However, I think that the strongest argument is the fact that na is an all-purpose preposition, which is another way of saying preposition par excellence. According to the discussion around Figure 10, na could, in principle, travel in all directions from the center of Figure 7.

The third type of preposition-like construction I would like to discuss briefly in this connection are the ones illustrated in (3) above. Although Kihm (1994: 68) claims these constructions "are clearly fixed phrases, which means they must be specifically mentioned in the lexical entries for diyanti (sic!) and tras", I think that what is more important is the fact that here we have an additional use of the all-purpose preposition na. That is to say, si tras (his back) and si dianti (his front) are noun phrases that received a prepositional value by the simple addition of na. Indeed, this may have been the source of many spatial prepositions, as suggested elsewhere in this paper, namely, with the use of a noun indicating a part of space, or of the observer's body. In other words, some objects are placed behind (the side of the buttocks), in front of (the side of the face), inside (in the interior of the stomach, of the mouth), on the right/left hand side etc.

12. Are "complex prepositions" really complex prepositions or are they made up of simple prepositions?
Alain Kihm divides Kriol prepositions in simple and complex prepositions. Although, it is a practical division, it is hardly a unanimous one. As Kihm himself remarked, "all complex prepositions are made up of di preceded by an element which may be an adjective (lunju 'far', juntu 'close, similar'), an adverb (bas as in i sta la bas 's/he's over there below', dentru 'inside', disna 'long ago', filadu 'in front', fora 'outside', lungu 'alongside', riba 'above'), or a noun (banda 'edge', diyanti, tras)" (Kihm 1994: 67-68).

If we consider that, in fact, lunju and juntu are not adjectives per se, but adverbs, we can analyze the so-called "complex prepositions" as a construction containing an adverb with the scheme Adv+di, as is the case with lunju di, juntu di, bas di, dentru di, filadu di, fora di, lungu di, riba di and disna di. Those containing a noun belong to the scheme na+N+di. In the latter case, what we have in fact are the prepositions na and di. The first preposition (na) indicates location of something in the denotatum of the noun, whereas the second (di) indicates to whom or to which thing this denotatum belongs. Genitivity is one of the main functions of di.

The "complex prepositions" made up of adverb plus di would, at least apparently, present a different structure, namely, the already mentioned Adv+di. However, it is well-known since as early as the times of the Grammaire générale et raisonnée, by Arnauld and Lancelot, that adverbs are made up of a preposition and a noun. Thus, Latin sapienter (wisely) is an equivalent of cum sapientia (with wisdom), hodie (today) is, underlyingly, in hoc die (in this day). This means that Kriol li (here) corresponds, at least semantically, to in hoc loco (in this place). 

Based upon Arnauld's and Lancelot's idea, it is perfectly legitimate to consider dentru di, for instance, as equivalent to "in (the) interior of". If this analysis is valid, constructions that seem to have the structure Adv+di do have, in fact, the structure na+N+di. In short, all apparently "complex prepositions" turn out to be phrases that begin with na and end with di. This implies that all constructions of this type contain additional instances of these two simple prepositions. This explains why di occurs more than na in the statistics of Table 1. In several constructions, the idea of location indicated by na is conflated with the following noun. Plag (1998) discusses a similar case in Sranan.  

It is true that these constructions are relatively lexicalized to the extent that dentru di (inside of) is an alternative for the simple form na (inside of). However, there are constructions of this type that we may not consider prepositional. One example is na tempu di [fomi] (in the time of hunger/famine). We cannot say that this construction is a (complex) preposition. It is similar to na tempu di cuba (in the time of rain), na tempu di friu (in the time of cold) and so on. There are several phrases that have the same structure, like na kau di Jon (in John's place), na kau di sinta (on the place of sitting), na [tudu] koldadi di limaria (in [all]type of animal). The form na metadi di (between) means, literally, "in [the] middle of". There are several syntactic constructions that are parallel to those in question here. For example, kamalion pega na rabu di lubu (lit. "the chameleon grasped in [the] tail of [the] wolf"). What happens is that constructions like na metadi di occur more frequently than the latter one. Their higher frequency gives the impression that they are a unit. In summary, what we have here are the prepositions na and di.  

In English there are similar constructions, as  "in the heart of Paris" and "in the heart of the matter". French "au sein de" and Portuguese "no seio de" have the same meaning. I think that no one would consider them complex prepositions in these languages.    

13. Spatial prepositions and the environment of language
Since prepositions have traditionally been considered devoid of extralinguistic meaning, one might be led to think that they do not have anything to do with the Natural Environment of Language. In the present section I would like to discuss this topic in more detail. In fact, most, if not all, prepositions do have to do with the physical world, albeit indirectly, at least at the moment of their origin. 

It is implicit in Sapir's paper mentioned above that only a part of the lexicon has to do with the natural environment directly, namely, content words. This is the case with toponyms, anthroponyms, names of specimens of flora and fauna, among others. These names refer to things of our immediate environment. Functional categories like prepositions would not belong here. However, we have evidence that, at least from a genetic point of view, they, too, can be related to the world. In fact, all the relations shown in Figure 8 emerged and became stored in the minds of  community members  because in early times they had direct contact with things of the world that were somehow related to one another.

In order to avoid misunderstanding about the several uses of a lexeme of any language, it is advisable not to lose sight of the aspect of the environment it designates for the collectivity. This applies not only to lexemes that designate "things", but also to those that refer to relations between them. It is not enough to take into consideration only the context of the utterance in which prepositions occur. In essence this implies that not only the vocabulary may be related to the environment, but also some aspects of grammatical words and even grammar itself, as stated by Heraclitus and Haiman (1980).

The thesis that I defend here is that all prepositions have to do with the physical world. Some of them indicate relations that obtain between an observer A and something B. In this case, A may consider B lying "at the back" (tras di, antis di, na si tras), "in the front" (dianti di, na si dianti) and "inside" of  itself (na, dentru di). B may also be "at its side" (na si ladu), in general, since Kriol does not have equivalents for "to the right/left of". All these positions in relation to A correspond to physical places around B.   

A strong argument in favor of this relationship of prepositions to the physical environment is the fact that frenti, for instance, comes from Latin frons, fróntis 'face', via Portuguese. It is probable that, in its origin, tras (back) may have meant something like buttocks. In English, "back" means the back side of the body, but it is also part of the complex preposition "in the back of". Kriol constructions like na ña dianti (lit. "in my front") and na ña tras (lit. "in my back") indicate that the prepositional construction has a noun as its head and, this noun is the name of a place relative to the observer. Indeed, Portuguese has the expression "na minha frente", which is perfectly parallel to Kriol na ña dianti. There is no construction parallel to na ña tras. However, we can say "Ele falou mal de mim nas minhas costas" (lit. he criticized me in my back, i.e. in my absence). Another form is "por trás" (lit. from behind), which has roughly the same meaning as "nas minhas costas".

The lexicalization of expressions such as these may have to do with the survival of our ancestors. For example, it was certainly important for individuals to know whether a lion was "in the back of" or "in front of" them. It was important, too, to know whether the lion was "before" of "after" a tree, a small hill and so on. If it was "before, it was also "near" the observer (a dangerous situation). Otherwise it was "far" from her/him. This knowledge was literally of vital importance. In this case, it was also essential for the person in question whether s/he was "inside" or "outside" the cave. Even if a lion was not in the neighborhood, it was important to be "inside" if it was going to rain.

The complex preposition na metadi di has to do directly with the physical world. It indicates something C located at a point lying "in the middle of" or "at half of" the line that may be traced between A and B.

Kriol riba in riba di (on, over, on top of) comes from Portuguese "riba" which, in turn, comes from Latin  "ripa, ae" (‘margin', in general of a river). The riverbank/margin is always at a higher level than the flowing water. In Kriol, "to go down, descend" is also riba, i.e., to come/go from above. The equivalent of this in other languages, may have to do, albeit distantly, with the head, i.e., the superior part of the body, or the top of mountains or of trees.

As for na (in, on, at) and its equivalents in most European languages, its meaning is entirely opaque today. However, when we consider equivalent expressions such as dentru di (same meaning), we may speculate that far back in the past they may have come from something meaning "pit", the stone of a fruit. As we have seen in the Ecology of Spatial Relations, this is the most primitive spatial relation.

In some languages, the equivalent of na (in) is still entirely transparent. In some Bantu languages, like Luganda, the relation is indicated by something like "be-at". In Yoruba, the conjunction meaning "and" was, originally, "to join". In Amharic, this function is indicated by a verb meaning "to repeat" (Givón 1979: 262-264). In this connection, we could adduce verb serialization in some African and creole languages. According to Bickerton (1981), the existence of verb serialization in creoles takes place in order to supply the paucity of prepositions to indicate cases. In the Portuguese creole São Tomense, for example, we have constructions of the type seen in (20).

(20) E fa da ine (3PS, talk, give, 3PS) 'hei talked to himj'

The English translation of (20) hides the real meaning of the sentence. Literally, it says that somebody (E 'hei') talked and "gave" his talk to somebody else (ine 'hej'). In other words, the function indicated by the English directional preposition "to" is indicated in São Tomense by the "movement" verb "da" (to give). Maybe it is not pure chance we are here dealing with the dative case.         

The fact that Romance languages do not have an abstract form to designate "to the right/left of" is suggestive of the relative importance of the same for human species’ survival in their environment. Thus, in my Kriol data there is no occurrence of even complex forms such as "to the right of" and "to the left of". When necessary, Kriol speakers borrow them from Portuguese. One possibility could be *na direita di (< à direita de 'to the right of') and *na iskerda di (< à esquerda de 'to the left of') or na si ladu diritu (lit. 'at his/her right side') or na ladu di si mon diritu (lit. 'at the side of his/her right hand'). What in fact exists in Kriol is the general na ladu di 'at the side of', regardless of being the right or the left side.

In Tok Pisin, there is an interesting situation. There are "only two words which really correspond to English prepositions: long and bilong". The second of them "is used to translate possessive phrases", as in "has bilong papa" (my father's house), "papa bilong mi" (my father). Regarding 'long', it "is used mainly for spatial relationships between objects, and relationships where the connection is less intimate that is the case with bilong". For example, "haus bilong bus" (the house in the bush), "wok long gaten" (to work in the garden". "To express exact location, long is compounded with nouns indicating location: antap long 'on the top of'/above, ananit long 'underneath/below', arare long 'beside'", which remind us of the Kriol constructions discussed in section 12 (Laycock 1970: xxviii). An interesting fact is that the all-purpose preposition in Tok Pisin derives from the English verb "to belong". This word is composed of "be" plus "long", a spatial adjective. In other words, in their origin all prepositions may have been semantically transparent.  

Most Kriol prepositions are complex, namely transparent, in Sapir’s sense (1963). According to Sapir, "Where a transparent descriptive term is in use for a simple concept, it seems fair in most cases to conclude that the knowledge of the environmental element referred to is comparatively recent, or at any rate that the present naming has taken place at a comparatively recent time" (p. 16). There is the possibility of late borrowings from Portuguese. In this case, only those forms in the left column of Table 1 would belong to the ancient vocabulary of Kriol. However, this does not exclude the possibility that at least some of the complex prepositions in the second column of Table 1 may also be ancient.

14. Final remarks
My main purpose in this paper has been to relate prepositions to the Natural Environment of Language. In this sense, we have seen that the basic relationships are those indicated in the Ecology of Spatial relations. My conclusion has been that every relation contains in itself spatiality. Some are prototypically so, whilst others are nonprototypical. Although this is not all that can be said of prepositions, other aspects are not of concern here. It must be pointed out moreover that even spatial prepositions may cover a wide range of areas.

The all-purposeness of na is a case in point. If an observer is standing in the middle of a room, na may indicate all positions around her/him (above, under, at her/his side). Something may be na tetu (in the ceiling), na paredi (on the wall) and na con (on the ground). That is, even backness and frontness may be expressed by na if, for example, an object is on the back wall or on the front wall, respectively. Of course, if the observer is not referring to the walls, s/he may also use all the prepositions referring to the Ecology of Spatial Relations. This aspect of prepositions has been exploited by Vandeloise (1991) in detail.    

Arnault, Antoine & Claude Lancelot. 1830. Grammaire générale et raisonnée. Paris: Auguste Delalain. Reprinted in 1969 by Republications Paulet, organized by Michel Foucault.
Bärenfänger, Olaf. 2005. Merkmals- und Prototypensemantik: einige grundsätzliche Überlegungen. Linguistik online 12, 3/02 (access: 11/24/2005).
Bennett, David C. 1975. Spatial and temporal uses of English prepositions: An essay in stratificational semantics. London: Longman.
Bickerton, Derek. 1981. Roots of language. Ann Arbor: Karoma.
Calvet, Louis-Jean. 1999. Pour une écologie des langues du monde. Paris: Plon.
Coleman, Linda & Paul Kay. 1981. Prototype semantics: The English word lie. Language 57,1.26-44.
Couto, Hildo Honório do. 1973. Os conetivos. Universidade de São Paulo, M.A. thesis.
_______. 1994. O crioulo português da Guiné-Bissau. Hamburg: Helmut Buske Verlag.
_______. 2006. Ecolingüística: estudo das relações entre língua e meio ambiente. Brasília: Thesaurus Editora.
Døør,  Jørgen & Jørgen Chr. Bang. 1996. Language, ecology and truth - dialogue and dialectics. In: Fill (org.): 17-25. Fill, Alwin. 1993. Ökologie: Eine Einführung. Tübingen: Gunter Narr Verlag.
Eco, Umberto. 2002. A busca da língua perfeita. São Paulo: Editora da Universidade Sagrado Coração, 2nd ed.
Fill, Alwin. 1993. Ökolinguistik: Eine Einführung. Tübingen: Gunter Narr Verlag.
_______ & Peter Mühlhäusler (eds.) 2001. The ecolinguistics reader. Londres: Continuum.
_______, Hermine Penz & Wilhelm Trampe (eds.) 2002. Colourful green ideas. Bern: Peter Lang.
França, Aniela Improta. 2005. Neurolingüística. Ciência hoje, vol. 36,21.20-25.
Givón, Talmy. 1979a. On understanding grammar. N. York: Academic Press.
Greimas, A. J. 1966. Sémantique structurale. Paris: Librairie Larousse.
Haiman, John. 1980. The iconicity of grammar: isomorphism and motivation. Language 56,3.515-540.
Haugen, Einar. 1972. The ecology of language. In: The ecology of language. Stanford: Stanford University Press, p. 325-339.
Helbig, Gerhard. 1975. Geschichte der neueren Sprachwisseschaft. Hamburg: Rowohlt, 2a. ed.
Kihm, Alain. 1994. Kriyol syntax. Amsterdam: John Benjamins.
Lamb, Sydney M. 2000. Neuro-cognitive structure in the interplay of language and thought. In: Pütz, Martin & Marjolijn H. Vespoor (orgs.) Explorations in linguistic  relativity. Amsterdam: Benjamins, p. 173-196.
Lang, Jürgen. 1991. Die französischen Präpositionen: Funktion und Bedeutung. Heidelberg: Carl Winter.
Laycock, Don. 1970. Material in New Guinea Pidgin (Coastal and Lowlands). Canberra: The Australian National University.
Lefebvre, Claire. 1998. Multifunctionality and variation among grammars: The case of determiner in Haitian and in Fongbe. Journal of pidgin and creole languages 13,1.93-150.  
Lyons, John. 1977. Semantics. Cambridge: The University of Cambridge Press, 2 vols.
Makkai, Adam. 1993. Ecolinguistics: ¿Toward a new **paradigm** for the science of language? London: Pinter Publishers.
Martin & Marjolijn H. Vespoor (orgs.) Explorations in linguistic  relativity. Amsterdam: Benjamins, p. 173-196.
Mandelbaum, David (org.) 1949. Selected Writings of Edward Sapir in language, culture and personality.  Berkeley: University of California Press, p. 89-103.
Mufwene, Salikoko. 2001. The ecology of language evolution. Cambridge: Cambridge University Press.
Mühlhäusler, Peter. 2003. Language of environment - Environment of language: A course in ecolinguistics. Londres: Battlebridge.
Naess, Arne. 1997. Heidegger, postmodern theory and deep ecology. The trumpeter 14,4.
Odum, Eugene P. 1971. Fundamentals of ecology. Philadelphia: W. B. Saunders Company, 3. ed.
Ogden, C. K. & I. A. Richards. 1972. O significado de significado. Rio de Janeiro: Zahar Editores.
Peirce, Charles Sanders. 1972. Semiótica e filosofia. São Paulo: Editora Cultrix.
Plag, Ingo. 1998. The syntax of some locative expressions in Sranan: Preposition, postposition, or noun? Journal of pidgin and creole languages 13,2.335-353.
Pottier, Bernard. 1962. Systématique des éléments de relation: étude de morphosyntaxe structurale romane. Paris: Librairie Klincksieck.
_______. 1969. Grammaire de l’ espagnol. Paris: Presse Universitaire de France.
Rosch, Eleanor. 1975. Cognitive representations of semantic categories. Journal of experimental psychology: General 104, 3.192-233.
Sapir, Edward. 1963. Language and environment. Selected writings, 89-103. Berkeley: University of California Press (reprinted in Fill & Mühlhäusler 2001, p. 13-23; I quote from this).
Schaff, Adam. 1974. Linguagem e conhecimento. Coimbra: Livraria Almedina.
Thompson, R. W. 1961. A note on some possible affinities between the creole dialects of the Old World and those of the New. Proceedings of the conference on creole language studies. Londres: MacMillan, p. 107-113.
Trampe, Wilhelm. 1990. Ökologische Linguistik. Opladen: Westdeutscher Verlag.
Vandeloise, Claude. 1991. Spatial prepositions: A case study from French. Chicago: The University of Chicago Press.
Wittgenstein, Ludwig. 1968. Tractatus logico-philosophicus. São Paulo: Cia. Editora Nacional & Editora da USP.
Zimmermann, Klaus. 2004. Ecología lingüística y planificación lingüística. In: Lluís i Vidal Folch, Ariana & Azucena Palacios (eds.) Lenguas vivas en América Latina. Barcelona: Institut Català de Cooperació Iberoamericana, p. 93-109.
_______. 2005a. Die Entstehung der Kreolsprachen: ein Versuch aus konstruktivistischer Sicht. In: Endruschat, Annette & Axel Schönberger (eds.) Portugiesische Kreolsprachen: Entstehung, Entwicklung, Ausbau und Verwendung. Frankfurt/Main: Domus Editoria Europaea, p. 11-44.
_______. 2005b. Interferenz, Transferenz und Sprachmischung: Prolegomena zu einer konstruktivistischen Theorie des Sprachkontakts. In: Bidese et al (eds.) Das Zimbrische zwischen Germanisch und Romanisch. Bochum: Brockmeyer, p. 3-23.

A revised version of this paper was published  as The ecology of spatial relations: The case of Kriol prepositions. In: Schrader-Kniffki, Martina & Laura Morgenthaler García (eds.) 2007. La Romania en interacción: entre historia, contacto y política. Frankfurt/Madrid: Iberoamericana/Vervuert, p. 479-514.