The present experiments demonstrate that children as young as five years old (M = 5;2) generalize beyond their input on the basis of minimal exposure to a novel argument structure construction. The novel construction that was used involved a non-English phrasal pattern: VN 1 N 2 , paired with a novel abstract meaning: N 2 approaches N 1 . At the same time, we find that children are keenly sensitive to the input: they show knowledge of the construction after a single day of exposure but this grows stronger after three days; also, children generalize more readily to new verbs when the input contains more than one verb.
Keywords:verb argument structures; artificial language learning; novel construction learning; statistical learning
Running head: NOVEL CONSTRUCTION LEARNING IN FIVE-YEAR-OLDS 3A characteristic property of natural languages is the systematic correlation between structural patterns and abstract semantic or information structure functions (Fillmore, 1968;Pinker, 1989;Landau & Gleitman, 1985;Grimshaw, 1990). Such correspondences in the domain of argument structure-encapsulated by the notion of argument structure constructions-provide the basic clause types of a language (Goldberg, 1995). For example, the English sentences Katie gave Jack the book and Poppy baked Henry a cake are both instances of the ditransitive construction-a common phrasal pattern involving a subject and two objects. The two sentences contain distinct words but both convey actual or intended transfer. Our knowledge of this abstract linking is evident in the fact that we can use the construction productively-i.e., it can be used with new lexical items that may or may not lexically encode the transfer meaning. For example, if asked what She mooped him something means, speakers are quite likely to guess that she gave him something (Ahrens, 1995;Goldberg, 1995). In fact, adults generally interpret utterances with novel verbs by attending to the semantics of the argument structure constructions involved (Kaschak & Glenberg, 2000;Goldwater & Markman, 2009;Kako 2006; Johnson & Goldberg, submitted a).At the same time, there is a question about whether young children are able to use argument structure constructions in the same way as adults. There is a great deal of evidence that children's early productions tend to avoid straying too far from their input. For example, when children younger than three hear a novel verb used intransitively, they are highly unlikely to productively transitivize it (Akhtar & Tomasello, 1997;Baker, 1979;Bates & MacWhinney, 1987;Braine, 1976;Pinker, 1989;Tomasello, 2000). Such experimental data, along with data from spontaneous production (Bowerman, 1982;Tomasello, 1992;Lieven, Pine, & Baldwin, 1997;Ingram & Thompson, 1996), have led to the proposal that early grammars lack abstract argument structure representations and that apparent uses of a construction actually rely on verb-specific representations (so called verb-islands;Tomasello, 2000).Evidence from comprehension is somewhat more mixed. Experiment...