This paper investigates the influence of contrastive stress in resolving potential semantic ambiguities. The sentences under investigation contain the focus operator only. Sentences with only have three main properties: (a) some sentential element is typically in focus, (b) the speaker presupposes that a set of alternatives to the focus element (the contrast set) has previously been introduced in the context; and (c) the speaker makes the assertion that the focus element has some unique property which other members of the reference (...) set lack. Previous literature on focus has concluded that stress plays a role in the interpretation of only-sentences in many natural languages, including English (Rooth, 1992). In processing terms, this suggests that contrastive stress could immediately determine which NP bears the semantic focus (and therefore the NP that presupposes a contrast set in the context). On this scenario, the processing cost associated with contrastive stress should be no greater than the cost associated with neutral stress. There is, however, a theoretically motivated proposal that makes the opposite prediction (Reinhart, 1999). As we understand it, in this model the parser is expected calculate default stress first, with its associated interpretation, and only later reanalyze the sentence, if necessary, based on the contrastive stress pattern. Since in English the default stress falls on the last NP of the sentence, any deviation from this pattern will be more costly. In processing sentences with contrastive stress, the model implies that the parser first calculates the interpretation that is associated with the default stress pattern. In this regard, young children’s interpretation of sentences with stress and the focus operator only is of special interest. Because young children are in the throes of grammar formation, their behavior can reveal default strategies for sentence processing, as well as the extent to which the parser’s sensitivity to contrastive stress is universal, i.e., includes child languages.. (shrink)