David Barner

David Barner of the UCSD Psychology Department will speak at the UCSD Linguistics Department Colloquium on May 18, 2009, at 4:00 pm in AP&M 4301. (Note the later time!)

Meaning and verification in language acquisition

Humans, unlike other animals, acquire linguistic representations of objects, sets, and exact cardinalities (e.g., twenty-three). Also, we alone acquire advanced knowledge of mathematics, including explicit computations of addition, subtraction, multiplication, division, etc. Still, we share with other animals many cognitive capacities, including non-linguistic systems that support numerical computation. Recently, some researchers have suggested that human mathematical abilities, including counting, may be constructed on the basis of these non-linguistic resources that are shared with other animals - i.e., that these systems supply a hypothesis space for learning mathematical concepts. Others have suggested that learning to count requires special innate representations of exact number. Against both hypotheses, I will show that exact numerical concepts like "three" cannot be acquired from non-linguistic systems, and are not innate, but instead derive from logical representations provided by natural language. Number words, on this account, are initially acquired like other quantifiers, using the same resources. Non-linguistic number systems, act as Kantian "empirical intuitions" used to verify meaning hypotheses without actually defining them. In support of this, I will present evidence: (1) that non-linguistic systems lack the distinctions required by number words, (2) that quantifier acquisition is tightly linked to counting in development both within English, and cross-linguistically, and (3) that number words get their exact meanings via inferential processes normally associated with quantifiers. I will conclude that no special, innate, hypothesis space is required for acquiring number words in humans (as some have proposed), but that early integer knowledge arises from logical resources normally used to acquire quantifiers like "a", "some" and "all."