This talk reports the results of an experimental investigation of the acquisition of disjunction in Japanese. The English disjunction operator or yields a 'conjunctive' interpretation in negative sentences, as predicted by de Morgan's Laws. In contrast, the Japanese disjunction ka lacks the conjunctive interpretation within simple negative sentences. Our results show that Japanese children interpret ka conjunctively in simple negative sentences, in apparent disregard to the input they encounter. This suggests that the semantics of negation-disjunction expressions is governed by a parameter, which is initially misset by Japanese children.

Negated disjunctions in English and Japanese show the following contrast:

(1)  Taro didn't eat the carrot or the pepper.

      (=Taro didn't eat the carrot AND didn't eat the pepper)

(2)  Taro-wa  ninjin ka piman-wo  tabe-nakat-ta.

      Taro-TOP   carrot or pepper-ACC  eat-neg-past

In (1), English disjunction or is interpreted within the scope of local negation, yielding the conjunctive interpretation. However, this interpretation is not available for the Japanese counterpart, in (2). This cross-linguistic contrast can be captured by assuming that Japanese disjunction ka is a positive polarity item (PPI) (Szabolcsi 2002). Roughly, ka is like English some, taking wider scope than a local c-commanding negative expression.

Because Japanese disjunction ka must take scope above local negation, Japanese parents never use sentences like (2) in a situation where (1) is appropriate, i.e., when they know that Taro ate neither the carrot nor the pepper. For this reason, conjunctively interpreted disjunctions are significantly rarer in the input to Japanese children than for English-speaking children. If the acquisition of disjunction is fully input-driven, we should expect Japanese and English children to behave differently in interpreting negated disjunctions. It has already been shown that English children interpret the negated disjunction or conjunctively (Gualmini and Crain 2002). We examined whether Japanese children behave like Japanese adults, or like English children/adults.

In our experiments, we tested 30 Japanese children, using a Truth Value Judgment task. Children were asked to judge, for example, whether or not the sentence in (2) was true when Taro had eaten the carrot but not the pepper. In this situation the sentence is true under the adult Japanese 'disjunctive' interpretation of ka, but false under the conjunctive interpretation of ka. Whereas adult Japanese control group accepted the sentence 100% of the time in the above situation, children's acceptance rate in the same condition was only 25%. In short, Japanese children interpret ka conjunctively, as English-speaking children do. We independently confirmed that wide-scope, non-isomorphic interpretations are available to children (Lidz and Musolino 2002), in a control experiment using nanika (something). Children correctly accepted the test sentences with the PPI nanika 90% of the time. Therefore, children's narrow-scope, conjunctive interpretation of ka is not due to the unavailability of non-isomorphic interpretations. The fact that Japanese and English children manifest similar behavior suggests that a semantic parameter that is based on the PPI-hood of disjunction is operative in child language, and argues against input-driven approaches to language development.