Cognitive Aspects: How People Perceive, Memorize, Think and Talk About Landmarks - Landmarks: GIScience for Intelligent Services - page 76

conceptualize space differently, which shows in various language structures. Also
Mark et al. (e.g., [ 136 , 137 ] ) have demonstrated cultural differences between the
conceptualizations of geographic forms.
In this respect, spatial language can only be investigated in a language-conscious
manner; see for example the careful limitation to English in the work of Her-
skovits [ 81 ] and many others in the field, or Mark's assertion of differences between
cross-cultural conceptualizations, and correspondingly, cross-linguistic terms [ 134 ] .
Talm y [ 208 ] identified strong Gestalt properties in the spatial language of
localizing an object. He found that in an utterance conveying spatial information the
object to be localized, the figure , is typically smaller, or more movable or variable, or
more salient. The figure is localized by references to known, larger, more complex
and more stable objects, the ground . This way language structures space for the
mental spatial representation of the recipient. Tversky and Lee responded to Talmy
by pointing out that not only language structures space but also space structures
language, this means, whichever language is chosen, graphical or verbal, similar
kinds of information will be omitted or retained [ 223 ] .
Language is also not precise, adding to the uncertainty of an already abstracting
mental spatial representation (see Sect. 3.3.2 ) . More complicated, the mental spatial
representation of the speaker, or even only the subset in working memory, is not
fully conveyed in language. Language is a different representation medium, hence,
a transformation from mental spatial representation to language is required. This
transformation is lossy. The mental spatial representation is configurational, and
thus close to visual imagery, which is hard (or lengthy) to capture in all details
in words, as the proverb knows: “A picture is worth 10,000 words” (see [ 115 ] for
some history and scientific evidence). This argument for information loss relates
also to the linearization that is part of the transformation. A configuration has
two- or three-dimensional extent, and language is sequential. And last but not
least, spatial language can further abstract from knowledge in a mental spatial
representation. Language has been shown to be mostly qualitative about relations,
while configurational knowledge may still represent quantities such as lengths
or proportions. A qualitative description, however, can only invoke a schema or
prototype in the recipient's mind, and quantities remain undefined [ 208 ] . Recipients
even seem to reduce this indeterminacy by choosing preferential interpretations,
specific interpretations that in the process of communication only get revised later
if they lead to contradictions [ 105 ] .
The recipient's mind, however, is not a blank sheet when receiving a (verbal or
graphical) spatial description. Instead the recipient has prior spatial knowledge, at
least of the procedural type. Recipients have now two tasks: matching a received
description with their prior spatial knowledge, and using their constructed mental
spatial representations for the intended purpose.
Communication is usually not a one-way activity. Klein, for example, has studied
the dialog between two people about a route. The recipient can come back to the
speaker, ask for more detail, or for confirmation [ 100 ] . Similarly, in an experiment
called Map Task [ 7 ] participants had a dialogue about a route (see gray box).
While the map task itself aimed at supporting research in speech recognition, it is