Colour Terms, Syntax and Bayes Modelling Acquisition and Evolution

摘要: This thesis investigates language acquisition and evolution, using the methodologies of Bayesian inference expression-induction modelling, making specific reference to colour term typology, syntactic acquisition. In order test Berlin Kay’s (1969) hypothesis that typological patterns observed in basic systems are produced by a process cultural evolution under influence universal aspects human neurophysiology, an model was created. Ten artificial people were simulated, each which computational agent. These could learn denotations generalizing from examples inference, resulting had prototype properties characteristic terms. Conversations between these people, they learned one-another, simulated over several generations, languages emerging at end simulation investigated. The proportion terms type correlated closely with equivalent frequencies found World Colour Survey, most emergent be placed on one evolutionary trajectories proposed Kay Maffi (1999). therefore demonstrates how can emerge as result learning biases acting period time. Further work applied minimum description length form modelling particular problem investigated dative alternation English. presents learnability paradox, because only some verbs alternate, but children typically do not receive reliable evidence indicating participate (Pinker, 1989). presented this took note frequency verb occurred subcategorization, so able infer subcategorizations conspicuously absent, presumably ungrammatical. Crucially, it also incorporated measure grammar complexity, preference for simpler grammars, more general grammars would unless there sufficient support incorporation restriction. correct both alternating non-alternating verbs, generalise allow novel appear constructions. When less data observed, overgeneralized alternation, is behaviour when subcategorizations. results demonstrate learnable, may necessary account Overall, suggest forms determined much greater extent learning, cumulative historical changes, than expected if correct.