Topicality and Complexity in the Acquisition of Norwegian Object Shift

Abstract

Abstract This article investigates the acquisition of object shift in Norwegian child language. We show that object shift is complex derivationally, distributionally, and referentially, and propose a new analysis in terms of IP-internal topicalization. The results of an elicited production study with 27 monolingual Norwegian-speaking children (ages 4;05–7;00) reveal a prolonged delay in the acquisition of object shift with topical individuated pronominal objects. At the same time, we observe target-like placement and prosodic marking of contrastive, possessive, and indefinite pronouns, which do not shift. In our account of the observed delay we refer to the complexity models proposed by Yang (2002, 2004, 2005, 2010) and Hudson Kam & Newport (2005, 2009). In light of Yang's productivity model we argue that not shifting object pronouns is the rule, and OS is the exception. In light of Hudson Kam & Newport's approach to complexity, we argue that the children perceive OS as an inconsistent operation. Notes 1From here onward, all examples throughout the article are from Norwegian unless otherwise indicated. 2Adverbs behave just like negation with respect to OS. We illustrate this phenomenon here with negation as that is what we used in our experimental design. 3Note that Norwegian OS is different from scrambling in languages like German and Dutch, e.g., in its dependence on verb movement and in applying only to pronominal objects. See among others Thráinsson 2001 for extensive discussion of the differences between OS and scrambling. 4Holmberg 1999 takes OS to be a postsyntactic operation occurring outside of narrow syntax, at PF, and thus allows object shift to take place after verb movement, thereby violating the Extension Condition. Fox & Pesetsky 2005, on the other hand, propose an approach based on phases and order preservation, where the order of the verb and the object within the VP phase has to be preserved. Nilsen 2003 takes Holmberg's Generalization to follow from phrasal verb movement of the VP, where the object is pied-piped along inside the VP. 5Several possible landing sites for the object in OS have been proposed: adjoined to I' in Holmberg 1986, SpecAgrOP in Chomsky 1993, SpecTP in Bobaljik & Jonas 1996, Bošković 2004, or the specifier of an IP-internal TopicP in Jayaseelan 2001 and Josefsson 2001, just to mention a few suggestions in the literature. 6See also the responses to Erteschik-Shir in the 2005 special issue of Theoretical Linguistics, such as, among others, Bobaljik 2005, Holmberg 2005, and Svenonius 2005. 7Moreover, OS of pronominal objects in Norwegian also does not have the same interpretive effect as in Dutch. However, see the literature on OS of DP objects in Icelandic, which appears to have a similar interpretive effect to that found in Dutch (e.g., CitationThráinsson 2001; CitationVikner 2006). 8All contexts are provided in Appendix A. 9As described in the previous section, the two age groups were not presented with exactly the same number of contexts. We refer the reader to Appendix C for a comparison of only the contexts that were the same for the younger and the older age groups. 10It should be noted that the younger children produced considerably more relevant responses in the contrastive/indefinite condition than the older children: 39 out of 66 vs. 55 out of 160, respectively. One of the reasons for this is that the older children tended to substitute ikke nåkka 'not anything' by ingenting 'nothing.' In fact, they did so 25 times all together out of the intended 160. On the other hand, the younger children never used this perfectly natural alternative. Unlike the younger children, the children in the older group also tended to provide an elaborate answer in these test conditions, often producing structures like Han ringer ikke til jenta, men til gutten. (M01 6;09,09) 'He does not call the girl, but the boy.' This is another reason why the older children provided fewer relevant responses in this condition than the younger children. 1In Table 2 and all subsequent tables, the asterisks denote nontarget-like words. 11Recall that possessives are among the pronominal object types that do not shift in the target language, irrespective of whether they are mono- or di-syllabic. We have not discussed why possessive pronouns do not undergo OS, but one possible account would argue that these elements are inherently contrastive. We leave this question aside here. However, note that in the test conditions used in the present study, the possessive pronouns were explicitly contrastive, hence capital letters are used here to signal this. 12There is of course variation within the group, with some children performing more target-like than others. Note, however, that this group-internal variation does not clearly correspond to the age of the children within this group in the sense that the older children within the group perform more target-like than the younger ones. This is verified in the table with individual results in Appendix B. 13We are grateful to Paula Fikkert and Bruce Morén-Duolljá for help with this. 14Nevertheless, we do not wish to categorically exclude the possibility that there is a language change underway without verifying this more definitively. In the near future, we would like to go back and test the older group of children again to see whether they still leave a certain proportion of object pronouns unshifted. 15We are grateful to Charles Yang for working the numbers out for us and for helpful discussions of the relevant issues. 16Most typically, the children would occasionally use VP-referring pronouns in contexts 16 and 18 (cf. Appendix A), which describe needing a toothbrush and eating fruit, respectively.