Revisiting donkey anaphora in Mandarin Chinese

In their article published in this journal, Pan and Jiang (2015) challenge the claims and proposals made in Cheng and Huang (1996) concerning both the distributional patterns and interpretive strategies for donkey anaphora in Mandarin conditional. They claim that all three types of conditionals (rúguǒ-, dōu- and bare conditionals) allow either a wh-phrase or a pronoun in the consequent clause, and that both the wh-phrase and the pronoun may be either unselectively bound or interpreted by the E-type strategy. We show that, except for an observation already mentioned and accommodated in Cheng and Huang’s (1996) analysis of rúguǒ-conditionals, their distributional claims are incorrect. It is also shown that the interpretative flexibility they propose is untenable, as it leaves a number of otherwise well-predicted properties unaccounted for.

The Bound Variable Hierarchy and Donkey Anaphora in Mandarin Chinese

Cheng and Huang (1996) argue that both unselective binding and E-type pronoun strategies are necessary for the interpretation of natural language sentences and claim that there exists a correspondence between two sentence types in Chinese and the two strategies, namely that the interpretation of the “wh … wh” construction (which they call “bare conditional”) employs the unselective binding strategy, while the ruguo ‘if’ and dou ‘all’ conditionals use the E-type pronoun strategy. They also suggest that there is a complementary distribution between bare conditionals and ruguo/dou conditionals in the sense that the latter allows all the NP forms, e.g. (empty) pronouns and definite NPs, except for wh-phrases in their consequent clauses, and can even have a consequent clause with no anaphoric NP in it, while the former permits only the same wh-phrase appearing in both the antecedent clause and the consequent clause. Although we agree with Cheng and Huang on the necessity of the two strategies in natural language interpretation, we see apparent exceptions to the correspondence between sentence types and interpretation strategies and the complementary distribution between wh-phrases and other NPs in bare conditionals and ruguo/dou conditionals. We think that the claimed correspondence and complementary distribution are the default or preferred patterns, or a special case of a more general picture, namely that (i) bare conditionals prefer the unselective binding strategy and the ruguo ‘if’ and dou ‘all’ conditionals, the E-type pronoun strategy; and (ii) wh-phrases are more suitable for being a bound variable, and pronouns are more suitable for being the E-type pronoun. This paper proposes a Bound Variable Hierarchy to help account for the distribution of wh-phrases and pronouns in Chinese conditionals and claims that any deviation from the preferred patterns will require additional contexts or accommodation.

Two kinds of definites in numeral classifier languages

Numeral classifier languages distinguish definite noun phrases licensed by uniqueness from those licensed by familiarity. Unique definites are expressed by bare nouns or null pronouns, while familiar definites are expressed by indexicals such as demonstrative descriptions or overt pronouns. This generalization parallels the observation by Schwarz (2009) that German distinguishes unique versus familiar or anaphoric definiteness in its article system. The difference between the two kinds of definites can be reduced to the presence of a semantic index in the case of familiar definites. As familiar definites occur in most E-type contexts, including donkey anaphora, and uniqueness definites are not possible in these contexts, these facts provide support to dynamic analyses of E-type anaphora and pose problems for uniqueness-based approaches, such as the theory of Elbourne (2013).

Indefinite Donkeys on Islands

In this paper I present a theory of indefinites which captures two of their natural properties: indefinites license donkey anaphora (Geach 1962) and they exhibit ‘specific’ readings in which they appear to scope out of scope islands. In various flavours of dynamic semantics (Kamp 1981, Heim 1982, Dekker (2004)), these properties can be captured to the detriment of compositionally. Other theories have employed more involved technical machinery like choice functions (Kratzer 1998, von Heusinger 2002), Hamblin-semantics (Kratzer & Shimoyama 2002) or independence friendly logics (Brasoveanu & Farkas 2011) to derive exceptional scope readings, but ignored donkey anaphora. Theories of E-type anaphora, on the other hand, generally do not consider exceptional scope readings (Heim 1990, Elbourne 2001). My own analysis combines insights from dynamic semantics with referential indexing in LF-semantics, resulting in a fully compositional, static system.

Indefinite Donkeys on Islands

In this paper I present a theory of indefinites which captures two of their natural properties: indefinites license donkey anaphora (Geach 1962) and they exhibit ‘specific’ readings in which they appear to scope out of scope islands. In various flavours of dynamic semantics (Kamp 1981, Heim 1982, Dekker (2004)), these properties can be captured to the detriment of compositionally. Other theories have employed more involved technical machinery like choice functions (Kratzer 1998, von Heusinger 2002), Hamblin-semantics (Kratzer & Shimoyama 2002) or independence friendly logics (Brasoveanu & Farkas 2011) to derive exceptional scope readings, but ignored donkey anaphora. Theories of E-type anaphora, on the other hand, generally do not consider exceptional scope readings (Heim 1990, Elbourne 2001). My own analysis combines insights from dynamic semantics with referential indexing in LF-semantics, resulting in a fully compositional, static system.

AN “i” FOR ANi: SINGULAR TERMS, UNIQUENESS, AND REFERENCE

There is an interesting logical/semantic issue with some mathematical languages and theories. In the language of (pure) complex analysis, the two square roots of −1 are indiscernible: anything true of one of them is true of the other. So how does the singular term ‘i’ manage to pick out a unique object? This is perhaps the most prominent example of the phenomenon, but there are some others. The issue is related to matters concerning the use of definite descriptions and singular pronouns, such as donkey anaphora and the problem of indistinguishable participants. Taking a cue from some work in linguistics and the philosophy of language, I suggest thatifunctions like a parameter in natural deduction systems. This may require some rethinking of the role of singular terms, at least in mathematical languages.