We present a category learning experiment in which subjects faced the knowledge selection problem, i.e., they needed to use their observations to determine which prior knowledge would be useful for learning. The issue of putting prior knowledge into neural network models is reviewed, and we present a new model which addresses the knowledge selection problem. This model gives a good account of the experimental results.

We present two experiments investigating the production of implicit constructions. Using a confederate scripting paradigm we find that after making an inference participants were more likely to subsequently produce an implicature. This effect occurred at a global and a local level and was unaffected by the perceived role of the conversational partner. Our findings demonstrate that the choice of whether to be implicit is determined by the activation levels of representations specific to implicatures and that inference and implications have overlapping processing representations.

Recent evidence indicates that children represent and learn multiple meanings of ambiguous words from early in development (e.g., mail letter, alphabetic letter). This raises the question of which naturalistic factors might allow young children to resolve lexical ambiguities. Previous research has shown that children’s processing of ambiguous words is facilitated by verb-related information. However, it is still unclear whether such facilitation comes from bottom-up (lexical associations) or top-down information sources (verb-event structures). In this study, we leveraged a large sense-annotated child-directed speech corpus to disentangle the effect of bottom-up lexical and top-down event structure cues. Preliminary results show that 4-year-olds might rely on verb-event structures when these are put in competition with lexical association. We discuss implications for theories of sentence parsing and word learning.

Children struggle to derive scalar implicatures. Initially this was thought to relate to a lack of cognitive resources required for the computation. More recently however, there has been a shift towards the alternatives (what a speaker could have said but did not). The argument is that children struggle to make the scalar implicature associated with some because they are unaware of its relationship with the stronger alternative all. We present a priming study that investigates this. We show that children’s implicatures can be primed equally by alternatives in quantifier and ad hoc expressions. This suggests that children are aware of the scalar relationship between some and all, even if they choose not to derive the implicature.

Frequency counters are computational mechanisms that track the frequency or probability of speech units. Such counters are idealizations which re-describe frequency effects in early word segmentation, not providing an underlying learning mechanism from which these effects arise. Previous work has shown that Implicit Chunking represents a plausible learning mechanism explaining infants’ sensitivity to statistical cues when segmenting small-scale artificial languages (French et al., 2011). However, no work has examined whether Implicit Chunking allows to segment naturalistic speech in a developmentally plausible way. Here, we show how a novel symbolic model of Implicit Chunking – CLASSIC-Utterance-Boundary - performs better or as well as previous frequency-based models (i.e., transitional probability, chunking) at predicting children’s word age of first production and a range of word-level characteristics of children’s vocabularies (word frequency, word length, neighborhood density, phonotactic probability). We suggest that explicit frequency counters are not necessary to explain infants’ speech segmentation in naturalistic settings.