Statistical regularities in vocabulary guide language acquisition in connectionist models and 15-20-month-olds.

Abstract

This research tested the hypothesis that young children's bias to generalize names for solid objects by shape is the product of statistical regularities among nouns in the early productive vocabulary. Data from a four-layer Hopfield network suggested that the statistical regularities in the early noun vocabulary are strong enough to create a shape-bias, and that the shape-bias is overgeneralized to non-solid stimuli. A second simulation suggested this overgeneralization is due to the dominance of names for shape-based categories in the early noun vocabulary. Two subsequent longitudinal experiments asked whether it is possible to create word learning biases in children. Fifteento twenty-month-old children were given intensive naming experiences with twelve noun categories typical of the types of categories children learn to name early. The children developed a precocious shape-bias that was overgeneralized to naming non-solid substances. Further, these children showed accelerated vocabulary development. Children taught an atypical set of nouns or no new nouns did not develop a shape-bias and did not show accelerated vocabulary development.

Children construct novel word meaning ad-hoc based on known words: Computational model of shape and material biases

“…In contrast, material bias results aren't so stable: two-and two-and-a-half-yearolds showed material bias (Soja et al, 1991); so did three-, four-, and five-year-olds (Dickinson, 1988); meanwhile, toddlers ranging from 17 to 33 months old never showed material bias (Samuelson & Smith, 1999); two-, two-and-a-half-, and four-year-olds and adults who spoke Japanese and two-year-olds who spoke English showed material bias, but two-and-a-half-, four-year-olds, and adults who spoke English didn't (Imai & Gentner, 1997); three-, four-, and five-year-olds didn't show material bias when given such syntax information as determiners (Subrahmanyam, Landau, & Gelman, 1999). And surprisingly, Samuelson (2002) reported that children showed extraordinary shape bias for nonsolid stimulus, which is the exact opposite case of material bias. Here, we call such opposite bias "overgeneralized shape bias" because it looks like the overgeneralization of shape bias to material bias.…”

“…Linda Smith and colleagues led LBA studies (Smith, 1995;Samuelson & Smith, 1999;Smith, Jones, Landau, Gershkoff-Stowe, & Samuelson, 2002;Samuelson, 2002;Colunga & Smith, 2005). They have explained word learning biases by associative learning.…”

“…(2.2.1) Solid stimuli (a) No shape bias during a small vocabulary (Samuelson & Smith, 1999) (b) Stable shape bias after a middle vocabulary (Samuelson & Smith, 1999) (2.2.2) Nonsolid stimuli (a) No material bias during a small vocabulary (Samuelson & Smith, 1999) (b) Overgeneralized shape bias during a small vocabulary (Samuelson, 2002) (c) Robust material bias after a sufficiently large vocabulary (Soja et al, 1991).…”

“…Samuelson (2002), Samuelson & Smith (1999) used the following procedures to examine shape and material biases. Experimenter prepares two stimulus sets: solid and nonsolid.…”

“…The second reason is existence of overgeneralization of shape bias to material bias. Samuelson (2002) reported that children extended novel nouns assigned to nonsolid substances to objects similar in shape to the named substance. The overgeneralization is not easy to explain if one assumes specific modules for those biases.…”