Nomi Erteschik-Shir (Ben Gurion University of the Negev, Israel): Canonical Information Structure and Dependencies

Abstract

I examine and compare the superiority effects in English, Danish, and Hebrew and show that the variation in the effect is due to the variation in “canonical information structures” - the proper alignment between syntax and information structure. Whereas English requires alignment of Subject-Topic and Predicate-Focus, Hebrew, for example, only requires that the topic precede the verb and that the focus follow it. Proper alignment is a requirement on dependencies. It follows that the superiority effect in English (1) is not exhibited in Hebrew (2).

(1) *What did who buy?

(2) ma kana mi?
what bought who

(3) *ma mi kana?
what who bought

(4) mi kana ma?
Who bought what

(2) requires a context in which the preverbal wh-phrase is d-linked and therefore qualifies as a topic. Such a context does not improve (3). Proper alignment is to be found in both (2) and (4): In both the preverbal wh-phrase is aligned with the topic and the post-verbal wh-phrase is aligned with the focus. V-2 is not a requirement on Hebrew word order so in principle (2) should also be licensed. But this word order does not allow for proper alignment and a superiority effect arises.
Superficial crosslinguistic differences in the manifestation of superiority effects are due to language-specific encoding of information structure allowing for an explanation of the variation in the effect across languages. This approach is particularly appropriate since by tying the effect to information structure it predicts the sensitivity of the effect to context.