By Steven Pinker, Timothy E. Moore, Irving Biederman, Stephen P. Schwartz, Anne Erreich, Judith Winzemer Mayer & Virginia Valian
Read Online or Download Cognition, Vol. 7, No. 3 PDF
Best nonfiction_2 books
The fine details of concentration
- Challenger 1 & 2: Spearhead of the British Royal Armoured Corps (Concord 7505)
- GIMP: When Life Deals You a Crappy Hand, You Can Fold -or You Can Play
- Speech by G. M. Malenkov, Chairman of the Council of Ministers of the U.S.S.R. at the session of the U.S.S.R. Supreme Soviet, August 8, 1953
- Hettangian through Aalenian (Jurassic) guide fossils and biostratigraphy, northern Yukon and adjacent Northwest Territories
Additional resources for Cognition, Vol. 7, No. 3
These heuristics serve LAS well, but only because the subset of English grammar and the subset of HAM structures that Anderson has chosen correspond al- Formal models of language learning 257 most feature for feature. For example, the grammatical rule that specifies that sentences consist of a noun phrase and a verb phrase corresponds to the breakdown of HAM propositions into a subject and a predicate; the grammatical rule that breaks the predicate phrase into a spatial preposition and a noun phrase corresponds to the breakdown of a HAM predicate into a relation and an object, and so on.
For the tree-fitting heuristic, this means that one of the propositions in the semantic structure is tagged as the “principal” one, and its node will be highest when the semantic tree is fitted onto the string of words. The nodes connected to this “root” node by one link are placed one level lower, followed by the nodes connected to the root by two links, and so on. Thus if the main propositiori concerns what the cat did to the mouse, the heuristic will fit the tree depicted in Figure 3(c) onto the string.
A similar overgeneralization problem follows from the fact that verbs with similar semantic representations have different case structures, that is, require different numbers and arrangements of noun phrases in the sentences in which they appear. Thus “give” might appear in a semantic structure with a subject and two objects, corresponding to the giver, gift, and recipient. Using this structure, LAS could build rules that parsed “Rockefeller gave Brown a million dollars”, with two consecutive noun phrases after the verb; it would also Formal models of language learning 259 construct rules to parse “Rockefeller gave a million dollars to Brown”, with a noun phrase and a prepositional phrase.
Cognition, Vol. 7, No. 3 by Steven Pinker, Timothy E. Moore, Irving Biederman, Stephen P. Schwartz, Anne Erreich, Judith Winzemer Mayer & Virginia Valian