On the role of informativeness in spatial language comprehension

Abstract

People need to know where objects are located in order to be able to interact with the world, and spatial language provides the main linguistic means of facilitating this. However, the information contained in the description about objects locations is not the only message conveyed; there is evidence in fact that people carry out inferences that go beyond the simple geometric relation specified (Coventry & Garrod, 2004; Tyler & Evans, 2003). People draw inferences about objects dynamic and objects interaction, and these information become critical for the apprehension of spatial language. Among the inferences people draw from spatial language the property of the converseness is particularly appealing; this principle states that given the description "A is above B" one can also infers "B is below A" (Leveit, 1984, 1996). Thus if the speaker says "the book is above the telephone" implicitly the listener also knows that the telephone is below the book. However this extra information does not necessary facilitate the apprehension of spatial descriptions. If it is true that inferences increase the amount of information the description conveys (Johnson-Laird & Byrne, 1991), it is also true that this "extra-information" can be a disadvantage. In fact the spatial preposition used in the description can end up in being ambiguous because it suits more than one interpretation: The consequence is a reduction of the informativeness (Bar-Hillel, 1964). Tyler and Evans (2003) called this inferential process Best Fit. Speakers choose the spatial preposition which offers the best fit between the conceptual spatial relation and the speaker's communicative needs. This principle can be considered a logical extension of the notion of relevance (Grice. 1975; Sperber & Wilson, 1986) and an integration for the Q-Principle (Asher & Lascarides, 2003; Levinson, 2000a) according to which speakers have the duty to avoid statements that are informationally weaker than their knowledge of the world allows. This dissertation explores whether the inferences people draw on spatial representations, in particular those based on the converseness principle (Levelt, 1996), will affect the process that drive the speaker to choose the most informative description, that is the description that best fit spatial relations and speaker needs (Tyler & Evans, 2003). Experiment 1 and 2 study whether converseness, tested by manipulating the orientation of the located object, affects the extent to which a spatial description based on the preposition over, under, above, below is regarded as a good description of those scenes. Experiment 3 shows that the acceptability for a projective spatial preposition is affected by the orientation of both the object presented in the scene. Experiment 4 and 5 replicate the results achieved in the previous experiments using polyoriented objects (Leek, 1998b) in order to exclude the possibility that the decrease of acceptability was due to the fact that one object was shown in a non-canonical orientation. Experiment 6, 7 and 8 will provide evidence that converseness generates ambiguous descriptions also with spatial prepositions such as in front of, behind, on the left and to the right. Finally Experiment 9 and 10 show that for proximity terms such as near and far informativeness is not that relevant, but rather it seems that people simply use contextual information to set a scale for their judgments.