Comparing MOSAIC and the variational learning model of the optional infinitive stage in early child language

Abstract

This paper compares MOSAIC and the Variational Learning Model (VLM) in terms of their ability to explain the level of finiteness marking in early child Dutch, English, Spanish, German and French. It is shown that both models are successful in explaining cross-linguistic variation in rates of Optional Infinitive (OI) errors, although both models underestimate the error rate in English. A second set of analyses shows strong lexical effects in the pattern of errors across all five languages studied. This finding is problematic for the Variational Learning Model and provides strong support for the notion that OI errors are incomplete compound finites as instantiated in MOSAIC.