Do-support is Difficult to Do: Evidence from Doubling Errors

Abstract

This study investigates the L1 acquisition phenomenon of “doubling errors”; “tense and/or agreement is incorrectly expressed twice–once on the ‘fronted’ auxiliary and once on the main verb” (O’Grady 1997:166). I evaluate three hypotheses—the Subject Auxiliary Inversion (SAI) hypothesis, the movement hypothesis, and the do-insertion hypothesis—concerning doubling errors. Through the examination of the occurrence of doubling errors in yes/no questions and negative declaratives, it is found that doubling errors occur far more frequently in do-contexts than in non-do-contexts. This asymmetry between the frequency of doubling errors in do-contexts and those in non-do-contexts is accounted for by the involvement of do-insertion (the do-insertion hypothesis), not by an un-adult-like question formation rule (the SAI hypothesis) nor by an un-adult-like movement rule (the movement hypothesis). It is argued that the underlying factor in doubling errors is the demands of do-insertion itself. This is presumably because do-insertion is a costly language-specific process and is a cross-linguistically marked process. This paper provides cross-linguistic evidence and theoretical support for the claim that do-insertion is marked.