Title

Semantic manifestations of the developing theory of mind

Date of Completion

January 2007

Keywords

Language, Linguistics|Psychology, Developmental|Psychology, Cognitive

Degree

Ph.D.

Abstract

I investigate a connection between ways in which children typically interpret quantified sentences and an aspect of cognition known as the Theory of Mind or ToM. I have looked at children's errors known as quantifier-spreading (q-spreading) and argued that these errors, previously regarded as semantic, can be analyzed as a consequence of children's immature ToM, whereby a weakness in their ability to infer what is relevant or salient to others causes children to misinterpret quantified sentences, semantics of which requires integrating contextual information. Once children's difficulties with reading vague contextual clues are taken into account, we can see that their knowledge of semantics of quantifiers is adult-like as early as can be tested.^ The effect of developing ToM on interpreting quantified sentences becomes visible during selection of values for covert quantifier domain restriction variables, particularly in the interpretation of indefinites. I hypothesize that visual asymmetry triggers "q-spreading" responses not because of children's deviant semantic form, in which the universal ranges over both the agent and theme NPs and consequently requires the denotations of the two to be 'symmetrical' (i.e. members of the agent-set to be exhaustively paired with those of the theme-set and vice versa). Instead, I suggest that contextual asymmetry (an unpaired individual with the property denoted by the indefinite) causes "q-spreading" because children use the asymmetry in the visual context as a pragmatic clue when deciding how to select domain restrictions for the indefinite. Thus, if the picture contains a number of boy/wagon pairs and a salient unpaired wagon (in case of the sentence "every boy is pulling a wagon"), children are likely to see its salience as an indication of its relevance for the speaker and construct the domain restriction for the indefinite to include exactly one object—the extra wagon). Consequently, the sentence receives the "wide scope" indefinite interpretation, false in the situation depicted in this scenario. This theory leads to certain predictions with respect to quantifier scope interactions, relationship between visual salience and error rate, and non-universally quantified sentences triggering q-spreading-like errors, which I test in the experiment reported in the dissertation.^