The U-shaped curve as a by-product of the third factor in overregularization by children

Abstract

In this paper, we review and reject two formal models of overregularisation (e.g. fazi “doed”, trazi “bringed”), widely attested in children’s production. Our main departures with these accounts are that a) both are dependent on children’s breaking of theoretical principles; b) no learning mechanism is ever discussed to justify this; and c) although each account describes a system that indeed produces overregularised forms, they fail to predict or capture the three stages of the U-shaped curve, being an explanation solely of its dipping section. In an attempt to remedy those perceived flaws, our approach keeps theoretical principles intact, maintaining parity with adult competence and addressing points a) and b). Furthermore, we integrate the chosen descriptive toolkit of Distributed Morphology and general cognitive biases/abilities in the form of Biberauer’s Minimise Maximise Means and Yang’s Tolerance Principle in order to explain the developmental path acquisition takes throughout the U-curve, thus addressing point c). Finally, we argue that our proposal is naturally compatible with the notion of grammar competition, which might explain why children do not categorically overregularise irregular verbs, but rather do so at a given rate.