Acquisition Of Wh-in-situ In Mandarin-speaking Children

Wh-in-situ phenomena are different in languages such as Iraqi Arabic, French, Hindi, Malay, Lebanese Arabic, Japanese, Korean and Chinese. In Mandarin Chinese, a wh-phrase can be licensed in-situ ambiguously in the following instances:(1) Single wh-question,(2) Multiple wh-question,(3) Long distance LF wh-extraction,(4) Donkey sentence,(5) Wh-as a negative polarity item. All these wh-in-situ expressions get represented as the operator-variable construction of (6) at LF, where the wh-phrase is semantically decomposed into operator(OP), variable(VAR)(x), restriction (RES), and a scope:(6)[CP(OP,VAR(x),RES) C [scope... x...]]Alongside with the abundant literature about the derivation of (6) and its conditions in wh-extraction languages and wh-in-situ languages, there are quite a number of studies on the acquisition of wh-ex-situ, a few on acquisition of wh-in-situ, but almost none on the acquisition of multiple wh-questions and donkey sentences in Mandarin Chinese.The present study attempts to provide some evidence from a large data of Mandarin-speaking children’s acquisition of wh-in-situ in the above instances and offer some tentative explanations. The relevant acquisition data are from a standardized test on698Mandarin-speaking children aging from2;06to7;11in a receptive task of question-with-picture. The test results can be summarized as follows: (A)50%of the children from2;06to3;05are syntactically aware of interrogative force on C[+Q];4;00up to4;11witness an increase up to80%;5;00up almost reach100%.(B) The children’s development in wh-in-situ is sensitive to the restrictions in the OP-VAR constructions. The [RES]s of THING and PERSON make wh-questions much easier than those with TIME, PLACE, INSTRUMENT, MANNER. The hardest one is REASON in why question.(C) The development is also sensitive to the number of variables or occurrences of the variables:multiple wh-questions and donkey sentences all involve more than one variable or occurrence of variables, and they are harder than single wh-questions.(D) It is also sensitive to the distance between Spec-CP and wh-in-situ as evidenced in the long distance wh-extraction in languages with overt wh-movement. More than40%children in2;06-4;05and20%in4;06-7;11have difficulty in understanding wh-question containing complex NP with the wh-phrase inside.(E) The development of wh-question and that of wh-as polarity items go hand in hand. These acquisition facts suggest that(F) Children are endowed with the apriori knowledge of:(a) sentence typing (including C[+WH]),(b) the logical concept of OP-VAR construction,(c) variable restrictions, hence (d) the lexical options of null-OP[WH] for wh-question, null-OP[(?)] for donkey pronominal, different from those in the lexicon of wh-ex languages as proposed in Jee-Youn Shin (2005). Like Crain’s (2012) claim about the innate nature of logical connectives, all these must part of UG, which need time to fix the lexical options or re-internalized on the basis of simple accessible data (Chomsky,2004). That’s why some of the younger children are slow and others are quick in identifying wh-questions and donkey sentences, and that’s why older children are averagely better than young children in almost all instances.(G) The acquisition time course seems to be regulated by computational complexity of OP-VAR configurations. For instance, the pair-list readings a multiple wh-question typically elicits depend on an OP{WH} binding more than two VARs within the same clause as in OP{WH}(x, y)(...x...y...), but an OP{WH} binding two occurrences of the same VAR in two conjoined clauses for the donkey pronoun interpretation:OP{WH}(x1,x2)((...x1...)(...x2...)). The computational complexity makes the acquisition of multiple wh-questions and donkey sentences a prolonged process. It cannot be attributed to the burden of working memory as is suggested in most of the acquisition studies. The fact that children behave better in the long distance LF wh-extraction out of NP island than multiple wh-questions and donkey sentences also lends support to the computational complexity account rather than working memory account.