Spatial reference
All languages have some means of indicating the spatial position of objects. In English, this is achieved through terms like left and right, and front and back: these words, which have counterparts in a wide range of European and other languages, use the planes of the body itself to identify different regions of space (it’s on the left/at the front). In fact, this type of spatial reference seems so basic that it is hard to conceive of any serious alternatives. As a result, the systems of spatial reference involved in terms like left, right, front and back have often been assumed to be innate, and hence universal. For the great Enlightenment philosopher Immanuel Kant (1724–1804), for instance, the concept of space is not derived empirically from any experiences, but is one of the innate mental ‘intuitions’ which the mind brings to the understanding of the world (see Kant 1998).
Cross-linguistic study does not bear out this expectation of universality. Instead, it turns out that there are other systems of spatial reference employed in the languages of the world than the one familiar from English and other European languages. The large-scale investigation undertaken by researchers from the Max Planck Institute for Psycholinguistics, reported by Pederson et al. (1998) and Levinson (2003), tested how speakers of different languages describe the spatial relations between simple objects.
Pairs of speakers sitting out of view of each other were each presented with an identical set of twelve photographs like those in Figure 11.3, as part of an experiment known as the ‘Men and Tree Game’. One speaker, the ‘director’, described a particular photograph, and the other, the ‘matcher’, had to guess, on the basis of the director’s description alone, which picture was meant. The photographs were of a sort which necessitated the use of the language’s spatial vocabulary; because director and matcher could not see each other or each other’s pictures, the matcher only had the director’s verbal descriptions to go on. Because of this arrangement, games of this type are an efficient way of eliciting spatial reference terms.
Experiments like these were conducted around the world on speakers of a broad range of mostly small-scale, traditional languages, but also in languages of non-traditional, largely urban societies, like Dutch and Japanese. These experiments explored the varying frames of reference used in languages for spatial location. A frame of reference is ‘the internally consistent system of projecting regions of space onto a figure-ground relationship in order to establish specification of location’ (Pederson et al. 1998: 571). English, for instance, uses left/right and front/back to divide space into different regions which can then be used to locate an object (the figure) with respect to a reference point (the ground): the man (figure) is to the left of the tree (ground), the tree (figure) is to the right of the man (ground), and so on. We will concentrate here on the frame of reference used for transverse (left–right) relations.
European languages including English, Japanese, and other languages have a relative frame of reference. This is a system of spatial identification which uses information about the bodily arrangement of a speech participant, often the speaker. Languages with relative frames of reference use spatial expressions with meanings like ‘in front of me/behind me’ and ‘to (my) left/right’. For instance, in describing picture 3.6 in Figure 11.3, a Japanese player of the Men and Tree game produced the following sentence:
Just as in English, the Japanese terms for ‘left’ and ‘right’ designate spatial regions which project out from the speaker’s own body. As a result, if the speaker changes position, the description of an object as on the left or right may also change: the frame of reference is ‘relative’ to the speaker’s location.
The relative reference frame is highly familiar and intuitive to an English or Japanese speaker. But it is not the only one. Some languages also contain an absolute frame of reference. This is a system of spatial location which does not depend on the position of a speech participant, but which is anchored instead in unchanging features of the geography, like uphill/downhill distinctions, or in the cardinal directions (north, south, east, west). In absolute frame of reference languages that use the cardinal directions, one does not say ‘the man is on the left’; instead, one says ‘the man is at the eastern/western/ northern/southern side’. Arrernte (Pama-Nyungan, Australia) is an example of a language with an absolute frame of reference, as exemplified in (27), also a description of picture 3.6:
As speakers of a relative frame of reference language like English, it is easy to mistake what is going on in a sentence like (27). English speakers can superimpose the four cardinal directions onto a figure like 3.6 as conventional markers of the top (‘north’), bottom (‘south’), left (‘west’) and right (‘east’) sides. To say ‘the man is on the east side’ can, for us, simply be equivalent to ‘the man is on the right-hand side’. As a result, ‘the man is on the east side’ could be an appropriate description of picture 3.6, with ‘east’ functioning simply as an alternative for ‘right’. In Arrernte, however, the cardinal direction terms are not used like this. Instead, east really means East, the direction of sunrise. For Arrernte speakers, the words for ‘north’, ‘south’, ‘east’ and ‘west’ aren’t able to be used as conventional substitutes for top, bottom, right and left; they are literal descriptions relating figures to absolute coordinates of the external world. The speaker of (27) describes the man-figure as ‘on the east side’ because it really was on the Eastern (sunrise) side of the picture, given the way the picture had been placed on the table during the experiment, and the table’s orientation in actual space. The ability of an Arrernte participant in the Men and Tree game to identify the correct image thus depends on their ability to orient themselves in space with respect to the cardinal directions. The fact that Arrernte speakers are typically able to do this is, to Westerners, a surprising ability. Given that both players in the Men and Tree game were seated facing the same way, cardinal points can be appealed to in order to precisely identify position.
The relative and absolute frames of reference often combine. About half of the languages investigated in the Men and Tree experiments use both frames of reference. English speakers, for example, occasionally use absolute frames of reference, as when they say that someone lives to the west of the bridge, or when they describe themselves as going further inland, or towards the coast. Some languages, however, only use one of the two: this is the case with Arrernte.
Apparently the least common frame of reference in the languages of the world is the intrinsic frame of reference. This system only makes reference to intrinsic features of figure and ground: ‘the man is at the side of the tree, the tree is at the chest/face/back of the man’ and so on. In the intrinsic frame of reference, there is no way of dividing space which is independent of the objects in it. In languages with other frames of reference, by contrast, it is possible to refer to regions of space without making any reference to objects: we can talk about the left side of the picture, or the eastern side of the picture, for example. These possibilities are not available in a language with only an intrinsic frame of reference. In the Men and Tree game, the only way of conveying the pictured spatial relations is by anchoring the descriptions in the man or the tree themselves: descriptors which are independent of these objects, like ‘left/right’ or ‘north/south’, are unavailable.
A language using an intrinsic frame of reference is Mopan (Mayan; Belize). Here is a typical example:
This was the instruction given by the director in the Men and tree game as a way of identifying photograph 3.3 in Figure 11.3 above. Notice that, as a matter of fact, there are actually two pictures which meet the description of the tree being at the man’s chest: 3.1 and 3.3. These pictures are mirror-reflections of each other, differing only in their transverse (left right) orientation: precisely the distinction that is not made in intrinsic frame of reference languages. As a result, speakers of this language consistently failed to differentiate pictures 3.1 and 3.3 in the Men and Tree game: when prompted to identify 3.3, they chose 3.1, and vice versa. Mopan provides no means for conveying this distinction.
What about the other pictures involving a left–right contrast, specifically 3.5 and 3.6? Given our description of Mopan as an intrinsic frame of reference language, it may come as a surprise to learn that it contains spatial terms corresponding in form to left and right, lef and rait. But these terms have a crucial difference in meaning from their English analogues. In English, left and right project regions of space relative to the speaker. Looking at picture 3.6, for instance, we would say ‘the bush is on the left [of the man]’ or ‘the man is on the right [of the bush]’. This left–right division is anchored in the speaker: our left is, of course, the man’s right. This English form of spatial reference is non-intrinsic: it depends on more than the inherent features of the reference objects, but invokes a set of coordinates which originate in a speaker external to the scene.
With this in mind, consider the Mopan description of picture 3.6:
What the speaker means is that the bush is on the man’s right: a correct description. This strikingly illustrates the difference between an intrinsic and a relative frame of reference. In a relative reference frame, left/right divisions are based on a participant in the speech situation, often the speaker. On their own, ‘the left’ and ‘the right’ of a picture refer to the speaker’s left and right, and if this is different from the hearer’s, further specification is necessary. In Mopan, by contrast, lef and rait refer to parts of the object, here, the man’s right side. The bush is on the right-hand side of the man, and this form of spatial identification is exactly parallel to the one quoted in (28) above. Rait is just like ‘chest’: it refers not to a generalized spatial region, but to an intrinsic part of one of the objects in the scene. ‘Man’s right-hand side’ would thus be a more accurate translation in this context.
These Mopan results serve as a reminder that notions like the left/right contrast which we take to be experientially basic and therefore likely to be present in all languages may not prove to be universal. Claims about what is and isn’t conceptually or semantically basic should not therefore be made without close cross-linguistic comparison.
