Response to Aoun and Li

Kuno, Takami, and Wu (1999) showed that Aoun and Li’s (1993) syntactic analysis of quantifier scope interpretation in English, Chinese, and Japanese makes incorrect predictions about the grammaticality and ambiguity of certain types of sentences. They proposed instead an expert system based on the interaction of syntactic and other principles that ranks the relative strengths of the potential scope interpretations of a given sentence. Aoun and Li (2000) replied to Kuno et al. with a number of criticisms, which are refuted here.

In Kuno, Takami and Wu (1999), we showed that Aoun and Li’s (1993) syntactic analysis of quantifier scope interpretations in English, Chinese, and Japanese is fatally flawed, by pointing out the following problems:

1. 1.

   1. a. Aoun and Li’s analysis incorrectly predicts that certain types of double object and topicalization sentences are unacceptable;

   2. b. there are ambiguous sentences that their analysis predicts to be unambiguous; and

   3. c. there are unambiguous sentences that their analysis predicts to be ambiguous.

We observed that while 1c makes their analysis insufficient, 1a and 1b make it untenable. Having thus rejected Aoun and Li’s syntactic analysis of quantifier scope interpretations in English, Chinese, and Japanese, we proposed an expert system that is free from these three problems. Demonstrating that the quantifier scope interpretations of a given sentence result from the interactions of various principles, some syntactic, others non-syntactic, we proposed an expert system that takes all these principles into consideration, and arrives at a composite opinion as to the relative strengths of the potential scope interpretations of a given sentence.

In their reply, Aoun and Li (2000) present three counterarguments:

1. 2.

   1. a. Kuno et al.’s claim 1a ‘results from some interpretations of Aoun and Li’s work that are not necessarily warranted’ (133);

   2. b. Kuno et al.’s claims 1b and 1c can be circumvented by taking into account idiosyncratic lexical properties of quantifiers such as each and number expressions; and

   3. c. There are serious problems with the characterization and execution of Kuno et al.’s expert system in accounting for quantifier scope phenomena.

As we will show, however, Aoun and Li’s counterargument 2a does not hold because it is based on assumptions contradictory to those explicitly stated in Aoun & Li 1993. Furthermore, if these new assumptions were allowed to replace those in Aoun & Li 1993, they would make Aoun and Li’s quantifier scope analysis even more problematic. Aoun and Li’s counterargument 2b does not hold, either, because it is not sufficient to consider only idiosyncratic properties of quantifiers, and many sentences that are problematic for Aoun and Li’s analysis are still left unaccounted for. Further, Aoun and Li’s counterargument 2c is based on unreasonable interpretations of Kuno et al. and/or on nitpicking. This argument does not seem to raise any serious problems for Kuno et al.’s expert system analysis of quantifier scope phenomena. Space limitation does not allow us to take up 2c in this reply.

1. Double object constructions

Aoun and Li (1993) claim that their analysis predicts that sentences of the double object construction such as 3 are unambiguous, and have only an ‘Indirect Object Q > Direct Object Q’ interpretation.

1. 3. John assigned one student every problem. (well-formed and unambiguous) They have hypothesized the S-structure 4b for sentences of this construction.

2. 4.

   1. a. They assigned Mary a tutor.

   2. b. S-structure hypothesized by Aoun and Li

Whether NP₂ (a tutor) c-commands NP₁ (Mary) in the above S-structure is a crucial issue in evaluating Aoun and Li’s quantifier scope analysis, because if it does, Aoun & Li 1993 incorrectly predicts that 3 will be unacceptable (because there is no well-formed LF-representation for the sentence in their framework). But if NP₂ doesn’t c-command NP₁, A & L’s analysis correctly predicts that the sentence will be acceptable, and unambiguous, with only a ‘one > every’ interpretation.

Aoun and Li (1993: 201, n. 8) have explicitly adopted Reinhart’s (1983) c-command notion, given in 5, that allows a one-step V* to V...