Alan Ke presented his research at the GLOW in Asia X conference over the summer. GLOW in Asia is one of the most important international meetings that focus on generative approaches to linguistics, and we are proud that our Department was represented at this meeting. Alan's research, conducted in collaboration with Ya Zhao, Liqun Gao, Shuying Liu, presents the first ever experimental evidence about the argument structure of so-called "relational nouns". The title and abstract of the presentation follow given below.

Null Anaphoric Possessor Arguments of Kinship Nouns and Long-Distance Binding in Mandarin

Researchers have noticed that relational nouns have an extra implicit argument inherent in their lexical meaning, but it is still unclear whether this argument is pronominal or anaphoric, and whether it is a theta assigned syntactic argument or only a semantic argument (Asudeh, 2005; Barker, 1995; Culicover & Jackendoff, 1995; Jackedoff, Maling, & Zaenen, 1993; Partee & Borschev, 2003; among others). To the best of our knowledge, no experimental evidence has been presented to directly address these questions. This study provides experimental evidence for a theta interpreted anaphoric possessor argument associated with kinship nouns, as a prototype of relational nouns, in Mandarin Chinese (MC). We also argue that there is a strong correlation between the interpretation of implicit possessors associated with kinship nouns and long-distance binding observed across languages.

We conducted two experiments indicating the syntactic presence of such an argument and the existence of a c-command requirement on its antecedent. A truth value judgment task paradigm (TVJT, Crain & Thornton, 1998) was adopted. 17 and 15 university students were tested in Experiments one and two, respectively. The experimenter told a puppet and the participants a story. After the story, the test sentences were presented by the puppet. The participants would judge the truth value of the sentences. Typical examples of test and control sentences in Experiment 1 and 2 follow. Each experiment consists of 4 test items and 4 control items.  

Experiment 1 

(1) a. test sentence:            Zhangsan dai-le          erzi qu Qingdao.
                                          Zhangsan bring-Asp  son  to Qingdao.
                                         ‘Zhangsan brought his son to Qingdao.’

     b. control sentence:      Zhangsan  dai-le           xiaogou qu Qingdao.
                                          Zhangsan  bring-ASP  dog         to Qingdao.
                                         ‘Zhangsan brought his dog to Qingdao.’

Experiment 2 

(2) a. test sentence:            Zhangsan zai Lisi zhuyuan     zhiqian dai     erzi  qu Qingdao.
                                          Zhangsan at Lisi   hospitalize before   bring son   to Qingdao.
                                          ‘Zhangsan brought his son to Qingdao before Lisi was hospitalized.’

      b. control sentence:      Zhangsan zai Lisi  zhuyuan    zhiqian  dai      xiaogou qu Qingdao.
                                          Zhangsan at  Lisi  hospitalize before    bring dog          to Qingdao.
                                          ‘Zhangsan brought his dog to Qingdao before Lisi was hospitalized.’

The test sentences employ kinship nouns while the control sentences do not. Note that in (2), Zhangsan c-commands the kinship noun erzi ‘son’ in (2a) and xiaogou ‘dog’ in (2b), while Lisi doesn’t. Given the context that Zhangsan brought Lisi’s son but not his own son to Qingdao, the participants consistently judge the test sentences as a false statement in both experiments over 88% of the time. However, when the kinship noun erzi “son” was replaced by a non-kinship noun such as xiaogou ‘dog’ (1b/2b), the participants consistently accepted the sentences as true characterizations of the context, over 91% of the time.

The striking difference supports our proposal that there is a null possessor in the syntactic representation of kinship, and that only nouns which c-command the kinship nouns can be taken as their antecedents. We therefore propose the structure of kinship nouns is [DP POSSESSOR [N]], where POSSESSOR is the extra argument of the kinship noun, and it can be null or overtly realized as a Possessive Phrase. When it is null, this extra argument is obligatorily bound by a c-commanding antecedent.

We also propose that this extra argument is anaphoric rather than pronominal: this hypothesis is based on at least four similarities we identify between the syntactic properties of this extra argument and the reflexive ziji ‘self’ in MC (Huang & Tang, 1991), such as C-command requirement, as shown in Experiment two, long distance binding, tendency to refer to the subject rather than the object, and blocking effect.

If the extra arguments of kinship nouns are anaphoric and their syntactic properties are similar to reflexives like ziji, this raises the possibility of a correlation in the occurrence of these two properties across languages. That is, if a language has bare kinship nouns that could be long-distance bound, it will also have a long-distance bound reflexive as a counterpart to ziji in MC. In the languages we investigated, we found evidence for a possible correlation between the existence of long-distance bound anaphoric possessor arguments of kinship nouns and long-distance bound reflexives, as summarized below: Italian, Dutch, Norwegian, Korean, Japanese, Russian have both long-distance bound reflexive (Cole, Hermon, & Sung, 1990; Progovac, 1993; Rappaport, 1986; Reinhart & Reuland, 1993; Sportiche, 1986) and long-distance bound null anaphoric possessor argument (Berns, 2008; Corver, 2007; Thunes, 2013), while Spanish and German have neither. The data concerning kinship nouns in Spanish, German are collected from consultants, and the same for the data concerning reflexive in Korean, Spanish and German.