Are infinitival to omission errors primed by prior discourse? The case of WANT constructions

Minna Kirjavainen, Anna Theakston

    Research output: Contribution to journalArticlepeer-review


    This paper examines the suggestion that infinitival to omission errors in English-speaking children can result from competition between two constructions (Kirjavainen et al. 2009a). Kirjavainen et al. suggested that the acquisition of two (or more) constructions (e.g., WANT-X and WANT-to) for verbs taking to-infinitival complement clauses can lead to infinitival to omissions, reflecting the relative frequencies of the constructions in the input. In the present study we analysed 13 English children's corpora to determine whether the presence of a variety of utterance types in the immediate discourse context preceding WANT-to-VP (e.g., I want to eat it) and erroneous *WANT-zero-VP (e.g., *I want - drink it) constructions was associated with infinitival to production/omission. This was done separately for the children's own and their interlocutors' discourse utterances. The data show that the occurrence of WANT-to and WANT-X constructions in the prior discourse was associated with differing proportions of infinitival to provision in the WANT-to/zero-VP construction. This together with Kirjavainen et al.'s (2009a) data suggests that children are learning at least two constructions for the verb WANT, that competition contributes to infinitival to omissions, and that the strength of competing representations is affected by overall input frequencies and the preceding discourse. © 2011 Walter de Gruyter GmbH & Co. KG, Berlin/Boston.
    Original languageEnglish
    Pages (from-to)629-657
    Number of pages28
    JournalCognitive Linguistics
    Issue number4
    Publication statusPublished - Nov 2011


    • Competition between constructions
    • corpus analysis
    • discourse
    • infinitival to
    • omission errors
    • priming
    • to-infinitive constructions


    Dive into the research topics of 'Are infinitival to omission errors primed by prior discourse? The case of WANT constructions'. Together they form a unique fingerprint.

    Cite this