How space structure language

B. Tversky and P. U. Lee. How Space Structure Language, volume Volume 1404/1998 of Lecture Notes in Computer Science, chapter Guest Contributions, pages 157–175. Springer-Verlag Heidelberg, Heidelberg, 1998.

The core of the paper is the “Schematization Similarity Conjecture”: To the extent that language and ‘ception (perception + conception) schematize things similarly, language will be successful at communicating space.

Essentially the authors argue that language is very good at encoding spatial features whereas is not equally efficient at encoding face features or emotional states. An example of this great ability to schematize space can be given by the English term ‘across’, where ideally the thing doing the crossing is smaller than the thing being crossed, and it is crossing in a straight path perpendicular to the length of the thing being crossed. Thus schematization entails information reduction, encoding certain features of the scene while ignoring others.

A second idea is that there seems to be levels of schematization that old over many context: people do not reinvent vocabulary and syntax at every encounter: space is decomposed into figures and the spatial relations among them, viewed from a particular perspective. Similarly, figures can be decomposed into their parts (Objects with features + frame). The experience of space is not abstract of empty space. Rather is the identity and the relative locations of things in space.

Language for figures selects one portion of a scene, as focal or primary, and describes it in relation to another portion, the ground, and sometimes in relation to a third portion of a scene. Figures are often reduced as points in space.

Objects have many identities but despite the possibilities, people are biased to identify objects at what has been called the ‘basic’ level: this is the level at which people seems to have the most information, indexed by attribute list, relative to the number of alternative categories that must be kept in mind. Objects are usually named by open-class terms. Shape seems to be the primary basis for categorization.

Some final ideas on spatial relations: figures are not just discerned and identified, they are also located. Figures are not located in an absolute way, but rather relative to other reference objects (other figures in the same scene) or reference frames (the set of figures surrounding).

In giving directions there seems to be a strategy in providing the minimal information for going from node to node. The schematization is more effective in depictions than description: although maps and directions schematize routes in similar ways, maps are more complete than directions, which need to be supplemented by inference rules.