The Weekly Newsletter of MIT Linguistics

Wu to talk in Goethe University Frankfurt Syntax Colloquium

Danfeng Wu will be giving an invited talk next Monday, November 23 (10:15 EST) at the Syntax Colloquium at Goethe University Frankfurt on “Syntax and prosody of either…or… sentences”.  Here is her abstract:

Prosodic structure largely reflects syntactic structure, but there are also mismatches. If we follow the intuition that prosodic structure is matched to pronounced material, an obvious place to study the syntax-prosody mismatch is syntactic structure involving non-pronounced material, such as ellipsis. In this talk I will present prosodic evidence of elided material in a phonetic experiment, where I show that the presence or absence of elided structure has an effect on the prosodic realization. Not only does this result provide a new source of evidence for ellipsis, but it also informs the question of what sort of syntactic information is accessible to prosody. 

The construction where I examine the prosodic effects of ellipsis is English either…or… coordination because it provides a suitable environment for the experiment, and allows me to design materials where ellipsis size could be parametrically varied. The prosodic work requires careful syntactic and semantic arguments that there is ellipsis in this coordination in the first place. As background, I will present evidence showing that there is ellipsis in either…or… coordination (following Schwarz 1999), and the size of the elided material is correlated with the position of either (as in 1a-d). These arguments rely on constituency tests, diagnostics involving elided pronouns and referring expressions, antecedent-contained deletion, and verb particle constructions.

(1) a. Lillian will look for either Lauren or Bella.
    b. Lillian will either look for Lauren or look for Bella.
    c. Lillian either will look for Lauren or will look for Bella.
    d. Either Lillian will look for Lauren or she will look for Bella.

After showing evidence for the analysis of ellipsis for (1a-d), I will move on to the prosodic part of the talk. The difference in ellipsis among (1a-d) might lead to a difference in prosody, specifically in phrasing. Consider (1d), which involves coordination of two clauses in syntax. If prosodic structure is built from a structure that contains elided material, and furthermore, if large syntactic constituents correspond to large prosodic constituents, we would have two large prosodic constituents, as can be observed by a large boundary (of intonational phrase, IP) between Lauren and or (2a). On the other hand, if prosody only considers surface structure, it might group Lauren or Bella as a single prosodic constituent even though they are not a constituent underlyingly, creating a small boundary (intermediate phrase, iP) between Lauren and or (2b).

(2) a. Either Lillian will look for Lauren IP) or she will look for Bella IP).
    b. Either Lillian will look for Lauren iP) or Bella IP).

If prosody is built from a structure containing hidden material, the boundary between Lauren and or would increase as we move from (1a) to (1d), since the amount of elided structure increases. In contrast, if prosody only considers surface structure, that boundary between Lauren and or would be the same for (1a-d). Preliminary results based on transcriptions of tones and breaks in the productions suggest speakers’ strong preference for (2a): the boundary between Lauren and or does increase as the elided material increases, suggesting that prosody tracks syntax closely. These results also bear on the question of timing: the non-pronunciation of material must occur after the creation of prosodic phrasing.