Tag Archives: coherence

Megan & Mike Fox Drink Beer & Talk about Linguistics. Do You Too?

I’m in the process of moving from Minnesota to Wisconsin & I’m just beginning a new full-time job so I won’t be having a lot of time to keep up with blog entries I’m afraid. Fortunately, I look forward to getting involved in various academic pursuits because I’m sure they’ll inspire great future entries. Additionally, I’m really kicking things into gear WRT applying to graduate programs. This all wouldn’t normally be too much for me to handle, but I still do get very worn out thanks to my still-healing body. I’m a bit worried that beginning a 40-hr/wk work schedule will exhaust me & put me in a lot of pain. Can you believe I’m not allowed to take aspirin for another 3 months? Anyway, enough about my personal life. Here’s an entry I wrote on ellipsis & coherence. — MLR

First, consider the following two examples:


a. Sandy walks and she chews gum.

b. Jerry does too, but not at the same time.


a. Sandy walks and Bill chews gum.

b. *Jerry does too, but not at the same time.

The site of VP ellipsis is marked by the stranded auxiliary does in both (1b) and (2b). In (1a), Sandy and she are most easily interpreted as being coindexed – the reason for this will be explained later – and the implied meaning of (1b) is that Jerry walks and chews gum, but he does not do both at the same time. The meaning of (2a) is straightforward, but (2b) is less easily interpretable; it is not clear whether Jerry walks and chews gum, or does only one or the other. Only but not at the same time necessitates that he do both, but this interpretation is not resolvable from the VP ellipsis site alone. For both (1) and (2), the VP ellipsis is resolved either at the syntactic or the semantic level of representation by borrowing something from its antecedent VP, but I will argue that a semantic approach better explains the VP ellipses.

A syntactic approach seems improbable for several reasons. A resolution of the VP ellipsis in (1b) at the syntactic level requires that an appropriate constituent be copied from (1a); however, walks and she chews gum is not a constituent structure and is therefore not replaceable by the auxiliary verb do. Similarly, walks and Bill chews gum of (2a) is not a syntactic constituent and cannot be pasted into the VP ellipsis of (2b).

A semantic explanation of the VP ellipsis in (1) and (2) seems to work much better. Returning to the assumption that Sandy and she are coindexed in (1a), it is easy to copy the semantic meaning from (1a) to (1b) to get a felicitous meaning. However, if Sandy and she do not refer to the same entity in the case of a sloppy interpretation, a problem arises, and (1b) is no longer felicitous; in fact, under these circumstances, (1) will resemble (2). For example, see (3):


a. Sandy walks and hei chews gum.

b. *Jerryj does too, but not at the same time.

It seems then that the infelicitous nature of (2b) stems from the ambiguity of having two entities in the antecedent phrase. That is to say, the VP ellipsis in (2b) and (3b) is irresolvable semantically because its antecedent causes ambiguity in attempting to establish the correct anaphoric reference. It is not clear what Jerry does too.

Both (1) and (2) involve resemble relations with parallel structures, which Kehler argues should implicate a syntactic parallelism at the VP ellipsis site. However, as we have seen, sentence (1b) is perfectly felicitous without having a syntactically parallels structure and the VP ellipsis is instead better understood semantically. Additionally, Frazier and Clifton argue that parallel clauses with contrastive objects as seen in (2b) do not “[capture] the effects of parallelism” (pg.337). To further illustrate this point, compare the following contrastive and non-contrastive parallel structures:


a. Sandy walks and she chews gum. (non-contrastive)

b. Sandy walks and Bill chews gum. (contrastive)

c. Sandy walks and chews gum (non-contrastive)

An interesting difference between the contrastive and non-contrastive sentences is that (4a) can be collapsed into (4c), whereas (4b) cannot be collapsed in the same way. This could be a potential explanation for why sentences like (2b) are infelicitous – they represent a semantic collapsing of a contrastive structure that is in its nature not collapsible into a single VP ellipsis. Therefore, Frazier and Clifton seem to have established a framework better suited to understanding VP ellipsis in examples like the ones examined here.

Megan L. Risdal



Filed under Linguistics